• Machine learning researchers taught a machine how to play Qbert for Atari.
• The computer program found a bizarre way to rack up 1 million points by playing the game "in what seems to be a random manner" and making the entire stage flash. 
• Artificial intelligence agents often find techniques to win games that a human would never discover. 

While the jury's still out on whether today's machine-learning techniques will ever create a program that could rival human intelligence, one thing about the future of artificial intelligence is clear: The machines are really good at playing games.

And when the machines get good at the games, sometimes they come up with bizarre strategies and tactics that a human never would. 

For example: In a new unreviewed paper posted on Arxiv, which we saw through a tweet from researcher Miles Brundage, three researchers from the University of Freiburg in Germany trained an agent using evolutionary strategies to play eight different Atari games from over 30 years ago. 

For one of the games, "Qbert," the AI found a way to exploit a bug in between levels, make the entire stage flash, and then rack up unlimited points.

Seriously. Even if you have never played "Qbert," you can tell that the agent is crushing the game. (The goal of the game is to visit every square in the level and make them change colors by jumping on them.)

Here's the video — the glitch starts at about 20 seconds in.

Here's how Patryk Chrabaszcz and the other researchers describe what the agent is doing in the paper: 

In the second interesting solution, the agent discovers an in-game bug. First, it completes the first level and then starts to jump from platform to platform in what seems to be a random manner. For a reason unknown to us, the game does not advance to the second round but the platforms start to blink and the agent quickly gains a huge amount of points (close to 1 million for our episode time limit). Interestingly, the policy network is not always able to exploit this in-game bug and 22/30 of the evaluation runs (same network weights but different initial environment conditions) yield a low score.

The strategies that AI agents take to win games are often fascinating. When Google's AlphaGo Zero agent beat the world's best Go player, its lead designer bragged that it found strategies that hadn't been used in the thousands of years the game has been played. "It found these human moves, it tried them, then ultimately it found something it prefers,” AlphaGo's lead programmer David Silver said at the time. 

It's also worth noting that the Qbert agent described in the new paper is using a different machine-learning technique from AlphaGo Zero's reinforcement learning. 

The bottom line is that machine-learning researchers love games. The rules are clear, you can run them thousands or millions of times, and they're just plain fun— even when the machines start breaking the games. 

Read the entire paper here. 

SEE ALSO: Watch a computer beat one of the world's best 'Super Smash Bros.' players

Join the conversation about this story »

NOW WATCH: How Silicon Valley's sexist 'bro culture' affects everyone — and how to fix it

• Google's newest hardware product, Clips, is a smart body camera designed to solve the problem of having to take out your smartphone every time you want to take a picture.
• The hope was that the built-in AI that lets the camera discretely snap moments you otherwise wouldn't have captured would make up for the $250 cost, but early reviews seem to say otherwise.
• Clips is a step in the right direction, but until someone comes up with a better device, a smartphone remains the best solution for capturing impromptu moments.

In October, Google announced an addition to its hardware lineup: a smart body camera called Google Clips. 

Clips was designed to solve some of the woes of the modern, social media-connected consumer: Clipped to yourself or set on a nearby surface, it would passively snap photos and videos, but only of moments its built-in A.I. deemed worthy of remembering.

This would ensure you get photographs of impromptu family moments without needing to hide your face behind a smartphone—helping you to stay more connected in the real world without losing the opportunity to share those moments online, too. At $249, Clips wouldn’t be cheap, but neither were some of its nearest competitors, such as GoPro’s $200 Hero5 Session.

The hope was that its internal smarts would make up for the cost, discreetly snapping delightful moments you’d never otherwise have captured.

Alas, early reviews of Google’s smart camera are now in, and that doesn’t seem to be the case. Because it clearly looks like a camera, it’s difficult for users and their families to relax and ignore the device’s presence—making it nearly impossible to capture the kind of candid moments promised. It’s overall photo quality isn’t great either, and since it can’t capture audio, the usefulness of its video capture is also limited.

In practice, the device’s smarts aren’t that helpful. It captures every moment in seven-second clips, which you then have to sort through in its accompanying app. In this app, you can choose a still from one of these clips to save as an image, or you can edit clips into a short video or GIF.

A.I. comes in more as the camera-app system learns what types of moments are important to you based on what clips you decide to save. To that end, it also learns the faces of family members and pets in order to surface “suggested clips” you may be more likely to save.

And despite housing a clip on the back, it’s not really designed to be worn as a body camera; it also needs to be moved around frequently—wherever the action is happening—in order to be effective.

“I’ve been testing the Clips with my two kids for the past couple of weeks, and…I can’t say I’m terribly impressed or happy with the results,” the Verge reviewer Dan Seifert wrote. “Most of the clips I’ve been able to capture didn’t look better or feel more authentic than what I’m already able to do with my phone or a dedicated camera.”

Engadget felt Clips’ auto-capturing capabilities were “too unpredictable” for the device to provide a totally satisfying experience. Several Clips reviewers felt the camera wasn’t worth the money.

In truth, Google is trying to enter a difficult space here, the holy grail of connected consumer photography: high-quality, memorable images with as little human involvement as possible. We’ve grown accustomed to capturing moments on the fly with our smartphones, but a growing contingent are trying to use phones more sparingly—or at least are aware of the issues they can cause with children.

A device like Clips would solve the conundrum for those who want to share photos of their life without constantly holding a smartphone in hand.

The reality is that Clips requires too much manual involvement to provide that solution, which begs the question: Would it be better off with a different form factor? Embedded into a pair of smart glasses, for example, you might escape that “there’s a camera in our midst” sense of unnaturalness—as long as smart glasses improve on their looks and adoption from the Google Glass era.

Or perhaps it’d do better to take the approach of other smart home devices. Jibo, an $899 smart home robot, is designed to similarly offer hands-free photography, albeit only when asked via voice. Other indoor security and home-monitoring cameras could do the job as well—the $229 PetCube, for example, specifically handles the issue of capturing video of your pet. It even dispenses treats to lure your pet into the frame. There has to be a toddler equivalent one day.

For most, a smartphone remains the best solution for capturing impromptu moments, even if it’s not perfect. It’s easy to carry with you, whip out as needed, and take high quality photos and videos.

There’s still room for a body or smart home camera to do this job better, but for now, Clips is a stepping stone. It’s a move toward a future where photography is more naturally integrated into our days without a gadget interrupting real-world experiences.

Unfortunately, we’re still reliant on that gadget—and the human involvement that comes with it.

SEE ALSO: This inconspicuous pair of glasses might just be the first fitness tracker I actually want to wear — here’s how it works

Join the conversation about this story »

NOW WATCH: Watch SpaceX launch a Tesla Roadster to Mars on the Falcon Heavy rocket — and why it matters

• Elon Musk didn't pull any punches when calling out Harvard professor Steven Pinker this week.
• Pinker said on a Wired podcast that "if Elon Musk was really serious about the AI threat he'd stop building those self-driving cars."
• Musk responded by saying that "if even Pinker doesn’t understand the difference between functional/narrow AI (eg. car) and general AI, when the latter *literally* has a million times more compute power and an open-ended utility function, humanity is in deep trouble."
  

Elon Musk took to Twitter this week to bemoan comments, made by Harvard professor Steven Pinker, that were dismissive of Musk's many warnings about the dangers of artificial intelligence.

Musk's smackdown came in response to statements Pinker made on a Wired podcast released a few days prior. At one point Pinker said that "if Elon Musk was really serious about the AI threat he'd stop building those self-driving cars, which are the first kind of advanced AI that we're going to see." He then pointed out that Musk isn't worried about his cars deciding to "make a beeline across sidewalks and parks, mowing people down".

Pinker's statement was off-the-cuff, but deeply silly on the face of it. Musk's frequent warnings about the dangers of artificial intelligence have touched on their possible use to control weapons or manipulate information to gin up global conflict. The physicist Steven Hawking, in a similar vein, has warned that AI could "supersede" humans when they exceed our biological intelligence.

Those concerns are about what's known as "general" artificial intelligence– systems able to replicate all human decision making. Systems like Tesla's Autopilot, the AlphaGo gaming bot, and Facebook's newsfeed algorithm are something quite different – "narrow" AI systems designed to handle one discrete task.

Developing Autopilot might make some contributions to developing general A.I., and it certainly raises moral questions about how we train machines. But, as Musk points out, even a completely perfect self-driving car would have nothing remotely resembling a fully enabled ‘mind' that could decide to either marginalize or hurt humans.

So the fact that a self-driving Tesla is specifically designed not to murder people isn't very relevant to the question of whether a machine intelligence might someday decide to. Further, it does nothing to dispense with the worries of organizations like the Electronic Frontier Foundation that limited A.I. systems designed for hacking and social manipulation are no more than a few years away.

Pinker, it must be pointed out, is not a technologist, but a psychologist and linguist. His early work developed the idea that language is an innate human ability forged by evolution – a position that itself has recently come under renewed scrutiny. Pinker's more recent work has praised science, reason, and the march of technology – but he might want to take a harder look at some of the risks.

SEE ALSO: Elon Musk posted a sensual Instagram picture with his tunnel-boring machine

Join the conversation about this story »

NOW WATCH: What would happen if humans tried to land on Jupiter

This story was delivered to BI Intelligence "Fintech Briefing" subscribers. To learn more and subscribe, please click here.

Wealth Wizards, a UK-based provider of digital financial advice services, has launched Turo, an AI tool that aims to provide access to faster financial advice software to financial advisory firms of all sizes, from startups to large incumbents.

The AI will tailor the insights it gives individual financial advisors to their company's advice policy and style by training itself on historical client interaction data. This will allow it to generate recommendations for advisors to give clients quickly, ensure the advice advisors give is in line with company policies, and boost case load efficiency by reducing the time spent on handling a case, allowing advisors to serve more clients, and drive revenue.

Wealth Wizards' approach to streamlining the process of giving financial advice via automation is unusual. The UK's Financial Conduct Authority (FCA) makes a clear distinction between "digital wealth management"— allowing software to invest your money for you — and "financial advice," and allows only the former to be conducted by computers.

When it comes to financial advice, for example, guidance on how to manage one's pension, the FCA prohibits delegation to software programs, and requires anyone dispensing recommendations to pass extensive exams. Wealth Wizards' strategy of using AI to generate insights for human advisors is a savvy way of bringing efficiency to the advice sector, as it maintains a human buffer between the AI and end clients.

While AI is still a nascent technology, it’s rapidly proving to be very capable at number-crunching large data volumes to produce actionable insights. As such, when applied to a sector that currently depends on humans processing increasingly unmanageable data troves to arrive at sensible conclusions, it promises to produce a lot of added value.

Given the prohibition of automating the dispensing of financial advice directly to clients, Turo will likely see high demand from financial advisors due to its ability to help advisors make more informed decisions and recommendations, where they may have done this more slowly or inaccurately without AI. This seems like a niche ripe to benefit from automation, and we will likely see more startups trying to capitalize on the opportunity, so Wealth Wizards is off to a good start by getting in early on the game.

Artificial intelligence (AI) is one of the most commonly referenced terms by financial institutions (FIs) and payments firms when describing their vision for the future of financial services. 

AI can be applied in almost every area of financial services, but the combination of its potential and complexity has made AI a buzzword, and led to its inclusion in many descriptions of new software, solutions, and systems.

Business Insider Intelligence, Business Insider's premium research service, has written a detailed report on AI in banking and payments that cuts through the hype to offer an overview of different types of AI, and where they have potential applications within the finance industry. In full, this report:

• Offers an overview of different types of AI and their applications in payments and banking. 
• Highlights which of these applications are most mature.
• Offers examples where FIs and payments firms are already using the technology. 
• Provides descriptions of vendors of different AI-based solutions that FIs may want to consider using.
• Gives recommendations of how FIs and payments firms should approach using the technology.

Subscribe to an All-Access pass to Business Insider Intelligence and gain immediate access to:

	This report and more than 250 other expertly researched reports
	Access to all future reports and daily newsletters
	Forecasts of new and emerging technologies in your industry
	And more!

Learn More

Purchase & download the full report from our research store

Join the conversation about this story »

• Elon Musk thinks AI is more dangerous than nuclear weapons.
• Speaking at the SXSW festival in Austin, Texas, Musk said that the rate of improvement in artificial intelligence "scares the hell" out of him.
• Musk warned that AI development should be regulated.

Elon Musk, the tech billionaire who wants to colonize Mars, is seriously worried about advances in artificial intelligence.

Speaking during a Q&A at the SXSW film festival and tech conference in Austin, Texas, on Sunday, Musk said the two things that stress him most in life right now are production difficulties with the Tesla Model 3 electric car and the dangers of AI.

"I'm really quite close, very close to the cutting edge in AI. It scares the hell out of me," Musk said. "It's capable of vastly more than almost anyone on Earth, and the rate of improvement is exponential."

Musk cited Google's AlphaGo, a software powered by AI that can play the ancient Chinese board game Go, as evidence of the rise of the machine. In early 2017, AlphaGo clinched a decisive win over the number-one player of Go, the world's most demanding strategy game.

Musk also predicted that advances in AI will let self-driving cars handle "all modes of driving" by the end of 2019. He said he thinks Tesla's Autopilot 2.0 will be "at least 100 to 200%" safer than human drivers within two years. Musk imagines drivers can sleep at the wheel someday.

The rate of improvement excites and worries Musk. He expressed a need for regulating AI development to ensure the safety of humanity, but he didn't say who should regulate it.

"I think the danger of AI is much bigger than the danger of nuclear warheads by a lot," Musk said. "Nobody would suggest we allow the world to just build nuclear warheads if they want, that would be insane. And mark my words: AI is far more dangerous than nukes."

Musk wants to create a Plan B society on Mars

Musk has a back-up plan in case nuclear war — or AI — wipes out the human race.

The SpaceX founder wants to put 1 million people on Mars as a sort of Plan B society. He told the crowd at SXSW that it would be ideal to get the base operational before a World War III-type event happens.

In the event of nuclear devastation, Musk said, "we want to make sure there's enough of a seed of civilization somewhere else to bring civilization back and perhaps shorten the length of the dark ages. I think that's why it's important to get a self-sustaining base, ideally on Mars, because it's more likely to survive than a moon base."

Musk has yet to detail exactly how hypothetical Mars colonists will survive for months or years on end. 

SEE ALSO: Elon Musk wants to colonize Mars with SpaceX — here's what he said it will be like as one of the first residents

Join the conversation about this story »

NOW WATCH: Elon Musk's The Boring Company sold out of these $500 flamethrowers

• Stack Overflow released its 2018 developer survey results, a massive look into the state of modern programming, with over 101,592 respondents.
• It finds that Google's Go programming language is a hair more "loved" than Apple's Swift, but that Microsoft's Visual Basic 6 is still the "most dreaded." 
• The survey also finds that developers are excited by the notion that artificial intelligence can automate jobs.
• The results also raise the alarming possibility that the tech industry is pushing out women and people from marginalized groups before their careers progress beyond the 5-year mark.

Today, the tremendously popular online programmer hangout Stack Overflow released its 2018 developer survey results, the absolute best look into the state of the software industry you can find anywhere.

With over 101,592 coders in 183 countries responding to the 30-minute survey — its biggest response yet — the Stack Overflow survey covers every conceivable topic, from trendy technologies to demographics.

Here are some key results, along with some insight from Stack Overflow CEO Joel Spolsky.

The Google-created Go programming language placed 5th in the "Most Loved" category, which measures how much developers want to keep working with a technology that they're already using. At a 65.6% approval rating, Go came in just ahead of Apple's Swift programming language, which clocked in at 6th place with 65.1%. Rust, a Mozilla-created programming language, was #1 with a bullet for the third year running with 78.9%.

The dubious "most dreaded" award

Also for the third year in a row, Microsoft's Visual Basic 6 won the dubious honor of "Most Dreaded," with a resounding 89.9% of developers currently using the language saying that they have no interest in continuing to do so. This was something of a personal affront to Spolsky, who describes VB6 to Business Insider as "the most perfect programming environment ever created." 

Along those same lines, the Python programming language is the "Most Wanted" technology, with 25.1% of developers who don't currently use it saying that they wish to acquire those skills. Spolsky ascribes Python's desirability to its popularity in the development of artificial intelligence — a field that the survey found was on the rise, with machine learning specialists and data scientists in the United States commanding an average salary of $102,000, the fourth-highest salary overall.

Otherwise, the survey gives some revealing detail into how software developers approach the intersection of technology and society. For instance, developers say that the most dangerous thing about artificial intelligence is the notion that algorithms might make important decisions, but that the most exciting aspect is that it could automate more jobs. 

9 out of 10 developers are men

On a final note, Spolsky points out one "sort of depressing" result of the survey: The largest single group of developers, accounting for 24.8% of respondents, have been working as developers for between 3 and 5 years. Another 11.4% have been working as developers for between zero and 2 years. 

There are two ways to look at that statistic: If you're an optimist, it means that the world of technology is fast-expanding, meaning lots of people finding new careers in programming. 

There's a more pessimistic reading, however. The results of the study show that 92.7% of respondents are male, and 74.2% of developers are white. Spolsky fears that those demographics, plus the prevalence of inexperienced developers, means that anybody who doesn't fit that profile tends to quit development by the 5-year mark.

"We push people out of the profession so quickly," laments Spolsky.

Read the full Stack Overflow survey here.

SEE ALSO: Google and the publisher of 'Assassin's Creed' are teaming up for a new weapon in the cloud wars

Join the conversation about this story »

NOW WATCH: Elon Musk’s artificial intelligence company created virtual robots that can sumo wrestle and play soccer

• Stephen Hawking died at age 76 on Wednesday.
• The world-renowned physicist worked on an Emmy Award-winning TV show called "Stephen Hawking's Favorite Places" before he passed away.
• In the show, Hawking flies around in a spaceship called the "S.S. Hawking" and explores his favorite cosmic mysteries.
• CuriosityStream released the final episode several weeks early and is streaming the three-part series for free for a limited time.

Stephen Hawking, who died today at age 76, was known for his work on the science of time travel and black holes.

The British physicist penned several bestselling books and even worked on an Emmy Award-winning documentary trilogy, called "Stephen Hawking's Favorite Places."

In the show, which is one of the last Hawking ever worked on, he flies around in a spaceship called the "S.S. Hawking" and explores deep scientific mysteries.

The show was created by CuriosityStream, and its description reads: "Mixing recollections from his childhood and family life that inspired his work as a scientist, he goes in search of the ultimate mystery: the theory of everything. Along the way, time travel and a precarious free fall to Venus, plus questions about aliens, God, and truth, offer unprecedented insight into this genius mind."

CuriosityStream planned to release the third and final episode, which in part dives into Hawking's fears about artificial intelligence, in mid-April.

But a representative for the company told Business Insider that, following the death of Hawking, its creators decided to release the last episode today.

Through March 23, Anyone can also watch the series for free for a limited time. It's normally packaged in a streaming subscription that costs between $2.99 and $11.99 per month.

You can find all of the "Stephen Hawking's Favorite Places" episodes at curiositystream.com/hawking.

Remembering Stephen Hawking:

• Stephen Hawking was my real-life Time Lord: Remembering the genius who inspired countless humans on this rock drifting through space
• A brief history of Stephen Hawking's time on planet Earth in one chart
• Stephen Hawking has died at 76 — here are some of the most remarkable and memorable things he ever said
• Stephen Hawking was only expected to live a few years after being diagnosed with ALS at age 21 — here's what the disease is
• Stephen Hawking died on a day that is cosmically connected to Albert Einstein and Pi
• Stephen Hawking had a 'sense of humor as vast as the universe': unique tributes flood in for esteemed scientist

Join the conversation about this story »

NOW WATCH: Stephen Hawking warned us about contacting aliens, but this astronomer says it's 'too late'

• Automated decision-making in machine learning can lead to discrimination.
• If this discrimination is not prevented, it would cause irreversible damages such as distrust of the technology and the companies that develop it.
• This is just one of the risks relating to machine learning.

Erica Kochi is the Head of Innovation at UNICEF and also leads the World Economic Forum’s Global Future Council

The opportunities that artificial intelligence (AI) can unlock for our world – from discovering cures to diseases that kill millions each year to significantly cutting carbon emissions – are expanding every day.  This includes a subset of AI called machine learning, which leverages the ability of machines to learn from vast quantities of data and use those lessons to make predictions. Machine learning (ML) is already enabling pathways to financial inclusion, citizen engagement, more affordable healthcare and many more vital systems and services. ML systems might highlight a post in your Facebook newsfeed based on your online activity, or select applicants in a hiring process. ML is one of the most powerful tools humanity has created – and it is more important than ever that we learn how to harness its power for good.

Public attention often focuses either on the existential threats artificial superintelligence poses to humanity (“the robots are coming to kill us”), or the opposite, salvation narrative (“AI will solve all our problems”). But there is a more immediate and less visible risk when machines make decisions: the potential reinforcement of systemic bias and discrimination. ML technologies are already making life-altering decisions for humans on a daily basis.  As Jim Dwyer wrote in the New York Times: “Algorithms can decide where kids go to school, how often garbage is picked up, which police precincts get the most officers, where building code inspections should be targeted, and even what metrics are used to rate a teacher.”

Learning not to discriminate

As we empower machines to make critical decisions about who can access vital opportunities, we need to prevent discriminatory outcomes. After all, machine learning is only a tool. The responsibility falls on people use it wisely – especially the people leading the way in its advancement, from corporate leaders down to system engineers. In other words, we need to design and use ML applications in a way that not only improves business efficiency but also promotes and protects human rights. Using technology to automate decisions isn’t a new practice. But the nature of ML technology – its ubiquitousness, complexity, exclusiveness and opaqueness – can amplify longstanding problems related to unequal access to opportunities. Not only can discriminatory outcomes in machine learning undermine human rights, they can also lead to the erosion of public trust in the companies using the technology. We must address these risks by evaluating the ways discrimination can enter ML systems, and then getting these systems to “learn” not to discriminate.

What happens when machines learn to discriminate?

Most of the stories we’ve heard about discrimination in machine learning come out of the United States and Europe. Events like a Google photo mechanism that mistakenly labeled an image of two black friends as gorillas, and predictive policing tools that have been shown to amplify racial bias, have received extensive and important media coverage. In many parts of the world, particularly in middle- and low-income countries, using ML to make decisions without taking adequate precautions to prevent discrimination is likely to have far-reaching, long-lasting and potentially irreversible consequences. Take, for instance, any one of the following examples:

• In Indonesia, economic development has unfolded unequally across geographical (and, subsequently, ethnic) lines. While access to higher education is relatively uniform across the country, the top 10 universities are all on the island of Java, and a large majority of the students who attend those universities are from Java. As firms hiring in white-collar sectors train ML systems to screen applicants based on factors like educational attainment status, they may systematically exclude those from poorer islands such as Papua.
• There are now ways for insurance companies to  predict an individual’s future health risks. Mexico is among the countries where, for most, quality healthcare is available only through private insurance. At least two private multinational insurance companies operating in Mexico are now using ML to maximize their efficiency and profitability, with potential implications for the human right to fair access to adequate healthcare. Imagine a scenario in which insurance companies use ML to mine data such as shopping history to recognize patterns associated with high-risk customers, and charge them more: the poorest and sickest would be least able to afford access to health services.
• While few details are publicly available, reports suggest that China is creating a model to score its citizens by analyzing a wide range of data, from banking, tax, professional and performance records to smartphones, e-commerce and social media information. The Washington Post described this as an attempt “to use the data to enforce a moral authority as designed by the Communist party”. What will it mean, in future, if governments act on scores computed using data that is incomplete or historically biased, using models not built for fairness?

These scenarios tell us that, while machine learning can hugely benefit this world, there are also important risks to consider. We need to look closely at the ways discrimination can creep into ML systems, and what companies can do to prevent this.

If, as Klaus Schwab argues in his book, The Fourth Industrial Revolution, we want to work together to “shape a future that works for all by putting people first, empowering them and constantly reminding ourselves that all of these new technologies are first and foremost tools made by people for people”, we need to design and use machine learning to prevent and not deepen discrimination.

This is an opinion column. The thoughts expressed are those of the author.

Read the original article on World Economic Forum. Copyright 2018.

SEE ALSO: So, what is machine learning anyways? Here's a quick breakdown

Join the conversation about this story »

NOW WATCH: What would happen if humans tried to land on Jupiter

Welcome to Digital Health Briefing, the newsletter providing the latest news, data, and insight on how digital technology is disrupting the healthcare ecosystem, produced by Business Insider Intelligence.

Sign up and receive Digital Health Briefing free to your inbox.

Have feedback? We'd like to hear from you. Write me at:lbeaver@businessinsider.com

CARDIOGRAM SHOWS HOW POPULAR WEARABLE DEVICES CAN DETECT AFIB: Popular wearables, like Apple Watch and Fitbit, can detect atrial fibrillation (AFib), or a common type of irregular heartbeat, according to a study published in JAMA Cardiology. The first large-N, peer-reviewed study was conducted by Cardiogram, a digital health company that uses machine learning to organize health data for preventative medicine, and the University of California San Francisco (UCSF). The results are significant because, despite the massive troves of wearable data that have been collected over the past few years, healthcare systems and wearable manufacturers have struggled to turn data into meaningful information — until now.

Researchers fed the data collected from 9,750 Cardiogram users enrolled in UCSF’s Health eHeart Study into the company’s AI algorithm, DeepHeart, to determine whether they had AFib. The neural network detected AFib at an accuracy rate of 97%, with sensitivity of 98% and specificity of 90%. DeepHeart is an artificial neural network — a form of AI — that uses heart rate and step count data collected by wearables to detect medical conditions.

Other companies are also exploring how to use wearables to detect heart conditions:

• AliveCor’s Kardiaband EKG reader received FDA approval in November 2017 as the first medical device accessory for the Apple Watch. The device attaches to a slot on the band of an Apple Watch, and the wearer can touch the sensor to get an EKG reading.
• Applepartnered with Stanford to run the Apple Watch Heart Study to monitor users’ heart rhythms using the Apple Watch heart rate sensor. Users experiencing AFib are alerted via an app and connected with a physician from telehealth company American Well.

Despite early progress, there’s still some way to go before these services achieve scale as medical detectors for AFib, according to co-author from UCSF Greg Marcus. Because participants had already been diagnosed with AFib, more testing needs to be conducted to determine how well the deep learning model will detect heart conditions in patients with no treatment history. Still, they’re showing promise as potential screening tools for heart arrhythmias. These heart conditions often go undiagnosed, as people typically don't experience outward symptoms. Early detection could reduce costly hospitalizations of patients due to preventable illnesses.

Enjoy reading this briefing? Sign up and receive Digital Health Briefing to your inbox.

MHEALTH APP PUBLISHERS ARE FALLING INTO OLD TRAPS: Fifty-six percent of publishers and developers looking to take advantage of the rapidly growing mHealth app market are making less than $10,000 annually, as they struggle to monetize their apps, according to a new report from Research2Guidance. The main issue is the reliance on old app industry measures and revenue models, such as volume of downloads, in-app advertisements, and premium content. Publishers that rely on these models tend to have the smallest share of apps generating more than $1 million in annual revenue. Meanwhile, just 11% of mHealth publishers bring in more than $1 million in revenue each year. Most of these publishers are focusing on licensing, service sales (remote consulting or coaching), and device sales (such as sensors).

Figuring out a revenue model early on is important for the longevity of an mHealth app or service, particularly given the ballooning costs associated with developing this content. On average, mHealth apps currently take between 15 months and two years to launch and cost around $425,000. That’s up 963% from the $40,000 it cost to develop an mHealth app in 2011. This is a lesson that developers in more established app industries, like gaming and e-commerce, learned early on. Moreover, it demonstrates the importance of finding proxies in related fields that provide an example of the best practices when entering a new business segment. What’s worked in similar, more established areas is likely to work in newer, growing segments.

EX-ATA CEO FORMS AI IN HEALTHCARE GROUP: Former American Telemedicine Association (ATA) CEO Jonathan Linkous has cofounded the Partnership for Artificial Intelligence (AI) and Automation in Healthcare, according to Politico. The group, which will include health systems, payers, regulators, and individual healthcare professionals, will strive to apply AI and automation to address the rapidly expanding costs, depleting human resources, and advancing technologies in healthcare. Collectively setting a standard could help these companies shape the future of AI in healthcare. Furthermore, banding together to explore AI and automation in healthcare research ensures that overall values and objectives are shared. Similar groups have been formed within other industries — with varying degrees of success — in order to facilitate and guide the development of AI and automation to benefit the broader industry. For instance, Google (and Google-owned DeepMind), Amazon, IBM, Microsoft, and Facebook together created the Partnership on Artificial Intelligence to Benefit People and Society (PAI).

MEDICAL ALARM MAKER INTEGRATES WITH AMAZON ECHO, ALEXA TO IMPROVE USAGE: Smart home device company TruSense has released a medical alarm pendant that can integrate with the Amazon Echo and Alexa, Amazon’s connected speaker and intelligence voice assistant, respectively, according to MobiHealthNews. The pendant is integrated with the Echo to allow users’ family members to check in with the device. Also known as personal emergency response systems (PERS), medical alarms are the most niche aspect of the telehealth industry and cater almost exclusively to elderly patients. Although the growing share of the elderly population means that demand for this segment of telehealth will grow, the increasing capabilities of remote patient monitoring devices, like fitness trackers and smartwatches, and mHealth apps will likely cause many of these designated devices to fold into the other segments of telehealth. Connecting these devices with emerging technologies like connected speakers and intelligent voice assistants to expand their practicality, while also making them easier to use and monitor, represents one way PERS companies can continue to grow in the evolving healthcare market.

Join the conversation about this story »

• Sixteen areas in China are using facial-recognition technology that can reportedly scan the country's population in one second, and the world's population in two seconds.
• Over the last two years the system has been used to arrest 2,000 people.
• Facial-recognition technology is soaring in China where it is being used to help consumers as well as police, who can track people's movements, friends, and even try to predict crime.

Across China, facial-recognition technology that can scan the country's entire population is being put to use. In some cases, the technology can perform the task in just one second.

Sixteen cities, municipalities, and provinces are using a frighteningly fast surveillance system that has an accuracy rate of 99.8%, Global Times reported over the weekend.

"The system is fast enough to scan China's population in just one second, and it takes two seconds to scan the world's population," the Times reported, citing local Chinese newspaper Worker's Daily.

The system is part of Skynet, a nationwide monitoring program launched in 2005 to increase the use and capabilities of surveillance cameras.

According to developers, this particular system works regardless of angle or lighting condition and over the last two years has led to the arrest of more than 2,000 people.

The use of facial-recognition technology is soaring in China where it is being used to increase efficiencies and improve policing. Cameras are used to catch jaywalkers, find fugitives, track people's regular hangouts, and even predict crime before it happens.

Currently, there are 170 million surveillance cameras in China and, by 2020, the country hopes to have 570 million — that's nearly one camera for every two citizens.

Facial recognition technology is just a small part of the artificial intelligence industry that China wants to pioneer.

According to a report by CB Insights, five times as many AI patents were applied for in China than the US in 2017.

And, for the first time, China’s AI scene gained more investment than that of the US last year. Of every available dollar going to AI startups around the world, nearly half went to companies in China.

SEE ALSO: China's 'Great Firewall' is taller than ever under 'president-for-life' Xi Jinping

Join the conversation about this story »

NOW WATCH: Facebook can still track you even if you delete your account — here's how to stop it

• Apple has hired John Giannandrea, who was Google's head of search.
• He'll work on machine learning and artificial intelligence and report directly to CEO Tim Cook, according to The New York Times.
• It's a big victory for Apple, as Giannandrea would be a hot commodity on the open market.

Apple has hired one of Silicon Valley's search and artificial-intelligence gurus to bolster its AI software, The New York Times reported on Tuesday.

John Giannandrea, Google's former chief of search, will join Apple and report directly to CEO Tim Cook, according to the report.

"John shares our commitment to privacy and our thoughtful approach as we make computers even smarter and more personal," Cook wrote in an internal email seen by The New York Times.

Giannandrea was one of the most powerful people at Google, overseeing the entire search organization. That means he has led teams that have used machine learning in products that billions of people have used.

That's the exact kind of expertise that Apple needs as Siri — its online assistant that's basically a search engine you access with your voice — increasingly faces competition from services like Google Assistant and Amazon Alexa. Apple's new smart speaker, HomePod, has also seen mixed reviews since launching earlier this year.

Machine-learning talent in Silicon Valley is in short supply, and qualified experts have been offered millions of dollars to jump from Google to other tech companies. So on Monday, when Giannandrea was said to have left Google, he immediately became one of the sought-after free agents in technology — but it looks as if Apple had him locked up the entire time.

Giannandrea does not have a lot of academic-style machine-learning papers in his name, The Wall Street Journal reports. Instead, he has focused on applying machine learning to production products.

But he is intensely respected in Silicon Valley, where he had stints at legendary '90s technology companies including Netscape and General Magic.

"He's an OG Valley technology visionary," the Netscape cofounder Marc Andreessen told Recode in 2016.

Giannandrea is "always on the leading edge," Andreessen said, adding: "Always learn new things when I talk to him."

Apple didn't immediately respond to a request for comment.

Join the conversation about this story »

NOW WATCH: A father and son are growing fruit and vegetables 8 metres below the surface of the Mediterranean Sea — here's why

I activated my Amazon Echo for the first time over two years ago, in December 2015. It's since become one of my favorite tech gadgets ever.

Amazon's family of Echo speakers are popular gifts right now, so many people are activating their Echo units for the first time.

Here's what you should know: These speakers, which can respond to either "Alexa,""Amazon," or even "Computer" (for those "Star Trek" fans out there), are quick to respond, and understand your commands far better than any other device I've used. Amazon Echo works exceedingly well wherever I am in my home. I can hear it — and it can hear me — almost perfectly.

Amazon Echo has completely transformed the way I live in my apartment. Here are my 19 favorite features:

SEE ALSO: Apple's 2016 report card: Grading all the new hardware Apple released this year

1. "Alexa, what time is it?"

Honestly, the best use cases for Amazon Echo are the simplest ones. With the Echo, I don't need to bother searching for my phone just to get the time — you can ask for the time from anywhere in your house and get the answer immediately. It's a small thing, but it totally makes a difference when you're rushing in the morning.

2. "Alexa, how's the weather outside today?"

Again, it's a simple task, but it's way quicker and better than pulling out your phone and opening your favorite weather app. Amazon Echo will not only tell you the current temperature, but also the expected high and low temperatures throughout the day, and other conditions such as clouds and rain.

3. "Alexa, set a timer for 10 minutes."

Amazon Echo is the perfect cooking or baking companion because it's totally hands-free. When the timer's up, a radar-like ping will sound until you say "Alexa, stop."

• Facebook CEO Mark Zuckerberg said that while it's relying increasingly on artificial intelligence to police content on its site, AI doesn't work well for identifying hate speech.
• AI won't be ready to reliably distinguish hate speech from legitimate expression for another five to 10 years, he said.
• Zuckerberg's comments came during his testimony Tuesday at a Senate hearing focusing on the Cambridge Analytica scandal.

Facebook is increasingly relying on artificial intelligence to identify content posted to its service that violates its policies, but CEO Mark Zuckerberg said there's one type of content AI struggles with — hate speech.

Indeed, it will take another five to 10 years for AI to be ready to police hate speech and be able to reliably distinguish it from legitimate political expression, Zuckerberg told senators during his testimony at a congressional hearing Tuesday. Although Facebook has worked on AI that could identify hate speech, the error rates are just too high, he said.

"We're not there yet," he said.

Hate speech is a problem for AI, because it's subject to lots of nuance, he said. Also, because Facebook operates in numerous countries around the world, its AI needs to understand those nuances in multiple languages.

"You have to understand what's a slur and whether something hateful," he said.

In his testimony, Zuckerberg noted that Facebook originally relied on its users to identify objectionable content. In the wake of the 2016 election and reports that Russian-linked actors hijacked Facebook's service to spread fake news and other propaganda, the company has been stepping up efforts to police content on the service. Facebook expects to have some 20,000 people working on security and reviewing content by the end of this year, Zuckerberg said.

More on Zuckerberg's blockbuster hearing:

• Business Insider's live blog of the entire hearing
• Zuckerberg hinted that Facebook could offer a paid version one day
• Zuckerberg says Mueller's office has interviewed Facebook employees in the Russia investigation
• Zuckerberg claims no knowledge of Palantir's involvement with the Facebook, Cambridge Analytica scandal
• Zuckerberg just renounced a core piece of Silicon Valley wisdom — and it could come back to bite Facebook
• Zuckerberg will not be under oath during his Senate hearing, but will still have a 'legal obligation to testify truthfully'

SEE ALSO: With Zuckerberg in the hot seat, here's what Congress should ask Facebook's CEO

Join the conversation about this story »

NOW WATCH: A Wall Street chief economist explains what could be the saving grace for mega-cap tech companies

• US Artificial intelligent startups raked in $1.9 billion in venture capital money in Q1 2018, up 29% from the quarter before. 
• AI accounted for 116 US deals, out of a total of 1,206 across tech sectors.
• AI accounted for three mega funding rounds — investments over $100 million — out of 34 across the US.
• Digital healthcare and cybersecurity also ranked alongside AI as the most notable Q1 funding trends.

Artificial intelligence startups in the US had a record quarter during the first three months of the year, raising $1.9 billion in venture capital across 116 deals, according to the 2018 Q1 MoneyTree Report published on Wednesday. That's 29% growth from the previous quarter. 

Digital healthcare companies, by contrast, raised $1.4 billion in Q1, while cybersecurity companies raised just $528 million. 

AI's strong Q1 numbers were underpinned by several mega deals in the space, including a $153 million series B for UiPath, a $112 million series A for Pony.ai, and a $100 million growth round for C3 IoT. 

The flow of money into AI startups comes as sophisticated computing technologies such as machine learning, speech recognition and image recognition have graduated from the labs and made their way into the commercial market. Many of these AI-based products have the potential to upend major industries, from automobiles to retail. 

All of the established tech giants — Google, Facebook, Amazon and Microsoft — are spending heavily to bolster their AI capabilities. But an explosion of small startups are also racing to develop new products and services that leverage artificial intelligence capabilities.

The AI startups come from Romania, China and Silicon Valley

UiPath is a Romania-grown and New York City-based "software robotics" company that uses AI to do digital busywork that humans don't necessarily want to do. UiPath's $153 million funding round, led by Accel Partners, capitalG and Kleiner Perkins Caufield & Byers, valued the startup at $1.1 billion, the company confirmed. 

Pony.ai, a Fremont, California and China-based self-driving car startup, raised its $112 million from Comcast Ventures, Legend Capital and Sequoia Capital China. 

C3 IoT, a Redwood City, California-based predictive analytics company, raised its $100 million from Sutter Hill Ventures, The Rise Fund and TPG Growth.

Those three mega funding rounds were among the 34 rounds over $100 million in the US in Q1, which accounted for 34% of the total VC funding over the quarter.  

Two of the three largest funding rounds in AI were early stage, following the trend of 10% deal growth for investments of that stage.

Earlier seed stage investments in AI — used to get companies off the ground — accounted for just 26% of all funding rounds, down from 35% in the quarter before.

Across industries, funding for all US-based, VC-backed startups increased 4%, up to $21.1 billion across 1,206 deals in Q1. But that number is supported by the larger deals, as the total number of investments actually declined by 2% over the quarter. 

SEE ALSO: It's not just the IPO market that's booming — here's why this Silicon Valley VC says 2018 will also be the year of M&A

Join the conversation about this story »

NOW WATCH: Why Russia is so involved in the Syrian Civil War

Welcome to Digital Health Briefing, the newsletter providing the latest news, data, and insight on how digital technology is disrupting the healthcare ecosystem, produced by Business Insider Intelligence.

Sign up and receive Digital Health Briefing free to your inbox.

Have feedback? We'd like to hear from you. Write me at: lbeaver@businessinsider.com

HEALTH PLAN LEADERS ARE MOST CONCERNED ABOUT MEMBER SATISFACTION: As healthcare information and options become increasingly available to consumers, health insurance leaders are reassessing their annual goals to better align with their customers’ needs. For the first time, the number one goal for US health plan leaders is improving customer satisfaction, according to a new HealthEdge survey. The majority of survey participants highlighted the need to modernize technology to achieve the goal of improving customer satisfaction. 

Online health insurer Oscar Health is a prime example of how payers use modern technology and a customer-centric approach to drive growth and retention. Oscar leans on its virtual services that offer convenience and encourage user engagement. In 2017, around two-thirds of Oscar’s customers’ interactions were virtual. Oscar’s Concierge team was accessed by 46% of Oscar members, its telemedicine service was used by 25% of customers, and the Care Router service — a yellow pages for finding physicians and booking appointments — was accessed by 45% of its members. These services aim to provide affordable care in place of more expensive healthcare services like emergency room visits.

However, cost structures are preventing many payers from innovating, according to HealthEdge. Nearly two-thirds of payer executives cited cost as the biggest barrier in the way of a product overhaul. Regardless, health plan leaders could find that they begin to fall behind the competition, unless they begin to make moves now to modernize their offerings. Because of this, we believe M&A activity will continue to pick up steam in 2018, particularly among larger players, such as Aetna and Cigna. New health tech innovations from companies like Amazon, Alphabet, and Apple could help to incentivize and drive the adoption of modern tech solutions as they explore entry points into the lucrative healthcare market.

Enjoy reading this briefing?  Sign up and receive Digital Health Briefing to your inbox.

FDA APPROVES FIRST AI-POWERED RETINOPATHY APP: The US Food and Drug Administration (FDA) greenlit the marketing of IDx-DR, the first medical device that uses artificial intelligence (AI) to detect diabetic retinopathy — a leading cause of blindness among people with diabetes. Developed by AI-diagnostics company IDx, the service uses AI to analyze images of a patient’s retinas to determine whether they have a form of retinopathy that requires a specialist examination. Because the software doesn’t need a specialist to run, it can be used by primary care providers, which vastly improves the availability of care. In turn, this could help increase the likelihood of catching diabetic retinopathy early. The FDA reviewed IDx-DR under its De Novo pre-market review pathway for low- to moderate-risk devices, which speeds up the process of new device approval. As the first AI product for diabetic diagnostics, IDx-DR will pave the way for more automated devices to enter the healthcare market. However, this isn’t the first AI-powered product to receive FDA approval. In February, the FDA approved the marketing of CDS software that alerts physicians of potential stroke in their patients.

AI as a clinical decision support (CDS) tool is gaining traction in the US. The technology shows potential for improving operational workflow by automating time-consuming tasks like image screening. And because it can be offered through primary care providers, it lessens the need to send patients on to emergency care or specialists. In addition, enabling primary care providers to see patients, rather than sending them off to specialists straight away, could help lower healthcare costs for payers. 

SURVEY PROVIDER AIMS TO STANDARDIZE TELEMEDICINE QUALITY OF CARE: As telemedicine adoption takes off among US providers, there's a growing need for a standardized test to determine the quality of care and patient satisfaction. To fill this need, leading US patient experience measurement and performance improvement firm Press Ganey developed two surveys that aim to measure a patient’s experience with virtual care, according to FierceHealthcare. Two surveys have been created — one measures fully virtual interactions, while the other measures partially virtual conference calls. The scientifically-validated questions address traditional things like the communication skills of the physician, but will also take into consideration how the telemedicine technology worked, whether it was effective, and how it impacted the patient experience. Creating a standardized test will also help physicians peer-review their scores, allowing expert opinion to counter consumer scores that don't necessarily reflect the physician's skill. As digital health offerings gain traction, there’s an opportunity for those that can back up the quality of their services to pull ahead of the competition. This will be particularly important as governments continue to ramp up regulations and incentives around telehealth solutions.

WEARABLES DATA ISN’T EXEMPT FROM DATA PRIVACY DISCUSSIONS: Data collected by wearables needs to be part of the ongoing data security discussion, according to leaders from Fitbit, UnitedHealth Group, and Empatica. Alongside security and privacy, health companies need to ensure transparency around how users’ data is being used and shared, Fitbit medical director John Moore said during a panel discussion at PULSE: The Atlantic Summit on Healthcare in Boston last week. The concerns discussed by industry leaders highlight the thin line that health organizations must tread. Continuous monitoring of users’ health has many benefits for a range of industries operating in health. For instance, it can help providers with remote patient monitoring, and give physicians a fuller picture of a patient’s lifestyle and behaviors. For insurers, wearables and health apps are spearheading preventative medicine, and could help lower costs associated with chronic illnesses. However, healthcare data is also being targeted by hackers. In early April, 150 million Under Armour-owned health app MyFitnessPal accounts were hacked, including usernames, passwords, and email addresses. If consumers become fearful of their data being accessed by nefarious players, it could mute the adoption and use of these tools. One way to combat this could involve regulation and policy that outline best practices for collecting consumer data, according to Empatica cofounder Rosalind Picard.

IN OTHER NEWS:

• Pharmaceutical giant Bayer commissioned a study that will explore how patients perceive digital products in healthcare, MobiHealthNews reports. The two-part pilot study aims to determine the feasibility of using digital products in future clinical trials. The move comes as demand for non-drug therapies gains momentum, threatening the pharmaceutical space.
• More than half of US consumers over the age of 40 would be comfortable with AI-assisted surgery, according to a new survey from SAS. That’s compared to just 40% of consumers under 40 who say the same. More so, consumers over 40 felt more comfortable with AI in health than they did with AI in banking or online retail. Overall, 47% of the 500 consumers surveyed say they’re okay with the idea of AI-assisted care.

Join the conversation about this story »

• At a conference this week, Microsoft Research scientist and leader Eric Horvitz says that the company has given up "significant sales" because it was worried the customer would use AI for not-good purposes. 
• Microsoft clarifies with Business Insider that the company has never cut off an existing contract over these concerns, but has turned customers away.
• Horvitz says that Microsoft has also placed contractual limits on what customers can do with AI, for ethical reasons.

Long-time Microsoft scientist Eric Horvitz says that the software company takes AI ethics so seriously, “significant sales have been cut off" because it was concerned that the potential customer would use its technology for no good. 

Horvitz, a director and technical fellow with Microsoft Research, made his remarks on stage at Carnegie Mellon University's K&L Gates Conference on Ethics and AI on Monday, as originally reported by GeekWire.

I got in touch with Microsoft for more clarity on Horvitz's remarks. The company confirmed that Microsoft had never cut off a deal with an existing customer — Horvitz was referring to the loss of possible revenue from potential customers. 

“Microsoft may decide to forego the pursuit of business proposals for numerous reasons, including the company’s commitment to upholding human rights," a spokesperson tells Business Insider. 

Beyond just cutting off those deals, Horvitz says that Microsoft has placed limitations on what customers can do with its AI tech: “And in other sales, various specific limitations were written down in terms of usage, including ‘may not use data-driven pattern recognition for use in face recognition or predictions of this type,'" he said, per GeekWire. 

That's an unusual point for Horvitz to make: Microsoft itself offers cloud-based services for developers to easily put facial recognition capabilities into their software. Still, Horvitz's remarks indicate that Microsoft is willing to place limits on what customers can and can't do with artificial intelligence. Microsoft declined to comment any further on that point.

"This committee has teeth"

In a more general sense, Horvitz was discussing Aether, an acronym for "AI and ethics in engineering and research," which is Microsoft's overall AI ethical oversight committee. “It’s been an intensive effort … and I’m happy to say that this committee has teeth,” Horvitz said.

“We believe it is very important to develop and deploy AI in a responsible, trusted and ethical manner. Microsoft created the Aether committee to identify, study and recommend policies, procedures, and best practices on questions, challenges, and opportunities coming to the fore on influences of AI on people and society," says a Microsoft spokesperson.

This approach is generally in line with Microsoft's public profile: The company's leadership has made much of the idea of ethical artificial intelligence, urging researchers, developers, and consumers to be responsible in how they use the technology.

"[We] want people to go forward in ways that are well informed, that are thoughtful, and in a sense, a commitment to shared responsibility. It is going to take a broad commitment to shared responsibility in order to ensure that AI is used well,"Microsoft President Brad Smith told Business Insider earlier this year. 

At the same time, the ethical use of AI is a hot topic in Silicon Valley: Earlier in April, Google employees petitioned the company's leadership to stop providing artificial intelligence to the military for use in drones. 

SEE ALSO: Thousands of Google employees asked CEO Sundar Pichai to stop providing AI tech for the US military's drones

Join the conversation about this story »

NOW WATCH: Humanity could be waste-free if we lived in a Circular Economy — here's how it would work

This is a preview of a research report from Business Insider Intelligence, Business Insider's premium research service. To learn more about Business Insider Intelligence, click here. 

Artificial intelligence (AI) is one of the most commonly referenced terms by financial institutions (FIs) and payments firms when describing their vision for the future of financial services. 

AI can be applied in almost every area of financial services, but the combination of its potential and complexity has made AI a buzzword, and led to its inclusion in many descriptions of new software, solutions, and systems.

This report from Business Insider Intelligence, Business Insider's premium research service, cuts through the hype to offer an overview of different types of AI, and where they have potential applications within banking and payments. It also emphasizes which applications are most mature, provides recommendations of how FIs should approach using the technology, and offers examples of where FIs and payments firms are already leveraging AI. The report draws on executive interviews Business Insider Intelligence conducted with leading financial services providers, such as Bank of America, Capital One, and Mastercard, as well as top AI vendors like Feedzai, Expert System, and Kasisto.

Here are some of the key takeaways:

• AI, or technologies that simulate human intelligence, is a trending topic in banking and payments circles. It comes in many different forms, and is lauded by many CEOs, CTOs, and strategy teams as their saving grace in a rapidly changing financial ecosystem.
• Banks are using AI on the front end to secure customer identities, mimic bank employees, deepen digital interactions, and engage customers across channels.
• Banks are also using AI on the back end to aid employees, automate processes, and preempt problems.
• In payments, AI is being used in fraud prevention and detection, anti-money laundering (AML), and to grow conversational payments volume.

 In full, the report:

• Offers an overview of different types of AI and their applications in payments and banking. 
• Highlights which of these applications are most mature.
• Offers examples where FIs and payments firms are already using the technology. 
• Provides descriptions of vendors of different AI-based solutions that FIs may want to consider using.
• Gives recommendations of how FIs and payments firms should approach using the technology.

Subscribe to an All-Access membership to Business Insider Intelligence and gain immediate access to:

	This report and more than 250 other expertly researched reports
	Access to all future reports and daily newsletters
	Forecasts of new and emerging technologies in your industry
	And more!

Learn More

Purchase & download the full report from our research store

Join the conversation about this story »

• Facebook is increasingly relying on artificial intelligence to identify offending items on its site.
• But AI is better at recognizing some things than others, CEO Mark Zuckerberg said on Wednesday.
• One example: it's good at finding nipples, he said. But not hate speech.

There's a growing fear of how artificial intelligence is going to affect society, but at least right now, AI is a lot better at some things than others.

One of the things it's pretty good at? Recognizing nipples.

That was the word from Facebook CEO Mark Zuckerberg on Wednesday. The social networking giant is increasingly relying on AI to police its service and identify content that violates its policies and guidelines.

But it's having varying degrees of success. Terrorism-related posts from the likes of ISIS and Al Qaeda are easy to  police with AI, he said. So too is nudity. But hate speech? Not so much.

"It's easier to build an AI system to detect a nipple than what is hate speech," Zuckerberg said on a conference call with analysts following the company's first-quarter earnings report.

Using AI, Facebook is able to identify and remove about 99% of terrorism-related content without a user having to notify the company first, he said. By contrast, it will likely take years for AI to reliably recognize hate speech he said. That difference is "frustrating," he acknowledged.

Zuckerberg's comments echo his testimony before Congress earlier this month during hearings examining the Cambridge Analytica scandal. He said then it could take five to 10 years for AI technology to mature enough that it would be able to reliably distinguish hateful slurs from legitimate political expression.

Meanwhile, his comparison to identifying nipples is an apt one. The company was embroiled in a controversy several years ago about people posting photos to the site of mothers breastfeeding their infants. After initially appearing to bar such photos, the company modified its terms of service to make clear that it would allow them — as long as no nipples are visible.

SEE ALSO: Mark Zuckerberg says AI won't be able to reliably detect hate speech for 'five to 10' years

Join the conversation about this story »

NOW WATCH: Here's the best smartphone camera you can buy

This is a preview of a research report from Business Insider Intelligence, Business Insider's premium research service. To learn more about Business Insider Intelligence, click here. 

Artificial intelligence (AI) is one of the most commonly referenced terms by financial institutions (FIs) and payments firms when describing their vision for the future of financial services. 

AI can be applied in almost every area of financial services, but the combination of its potential and complexity has made AI a buzzword, and led to its inclusion in many descriptions of new software, solutions, and systems.

This report from Business Insider Intelligence, Business Insider's premium research service, cuts through the hype to offer an overview of different types of AI, and where they have potential applications within banking and payments. It also emphasizes which applications are most mature, provides recommendations of how FIs should approach using the technology, and offers examples of where FIs and payments firms are already leveraging AI. The report draws on executive interviews Business Insider Intelligence conducted with leading financial services providers, such as Bank of America, Capital One, and Mastercard, as well as top AI vendors like Feedzai, Expert System, and Kasisto.

Here are some of the key takeaways:

• AI, or technologies that simulate human intelligence, is a trending topic in banking and payments circles. It comes in many different forms, and is lauded by many CEOs, CTOs, and strategy teams as their saving grace in a rapidly changing financial ecosystem.
• Banks are using AI on the front end to secure customer identities, mimic bank employees, deepen digital interactions, and engage customers across channels.
• Banks are also using AI on the back end to aid employees, automate processes, and preempt problems.
• In payments, AI is being used in fraud prevention and detection, anti-money laundering (AML), and to grow conversational payments volume.

 In full, the report:

• Offers an overview of different types of AI and their applications in payments and banking. 
• Highlights which of these applications are most mature.
• Offers examples where FIs and payments firms are already using the technology. 
• Provides descriptions of vendors of different AI-based solutions that FIs may want to consider using.
• Gives recommendations of how FIs and payments firms should approach using the technology.

Subscribe to an All-Access membership to Business Insider Intelligence and gain immediate access to:

	This report and more than 250 other expertly researched reports
	Access to all future reports and daily newsletters
	Forecasts of new and emerging technologies in your industry
	And more!

Learn More

Purchase & download the full report from our research store

Join the conversation about this story »

Welcome to Digital Health Briefing, the newsletter providing the latest news, data, and insight on how digital technology is disrupting the healthcare ecosystem, produced by Business Insider Intelligence.

Sign up and receive Digital Health Briefing free to your inbox.

Have feedback? We'd like to hear from you. Write me at: lbeaver@businessinsider.com

DOCTOR ON DEMAND SECURES $74 MILLION AMID TELEMEDICINE BOOM: Video-based telemedicine company Doctor On Demand secured $74 million in the latest funding round, led by Goldman Sachs and Princeville Global, bringing the company’s total funding to $160 million, according to TechCrunch. Doctor On Demand, one of the leading telemedicine companies in the US, plans on using the money to further develop its platform and expand the accessibility of its services. Telemedicine refers to the use of mobile and video technology to deliver clinical care, such as through patient-physician video conferencing. The startup is partnered with more than two dozen health services and serves more than 400 employer clients, bringing its reach to more than 2 million US patients.

The significant funding round comes as telemedicine balances on the verge of rapid adoption in the US. Business Insider Intelligence projects the US telehealth market — which includes telemedicine, mHealth, and remote patient monitoring — will grow at an annualized rate of 75% in the five years through 2023 to reach more than 57% of the total US population. This growth will be spurred by several drivers, including the following:

• A substantial addressable market. Although telehealth solutions aren't suitable for all patients, about 45% of the US population, or 147 million consumers, falls within the addressable market. Each year, there are more than 1.2 billion instances of outpatient medical care. Existing telehealth tools and solutions can serve anywhere between 42% and 45% of these visits using virtual care and RPM, equating to around 500 million instances that could be addressed by telehealth annually.
• The US government making telehealth a priority. No fewer than 12 pieces of legislation on telehealth were brought up in 2017, mHealthIntelligence notes. This was driven by the impending expiration of some bills, as well as the need to keep up with consumer and industry demand for telehealth technology.
• High demand from consumers. According to the 2018 Business Insider Intelligence Insurance Technology survey, 57% of consumers said they'd use telehealth for a remote general consultation if given the option. Further, 28% said they'd use telehealth if provided with a free in-person follow-up visit if it was required.

NOVARTIS GROWS ITS DIGITAL ROOTS FOR EYE-TEST DATA: On Wednesday, pharma giant Novartis launched FocalView, an iPhone app that will help researchers track ophthalmic disease progression without requiring patients to travel to their doctors, according to Reuters. Patients will self-report through the app by taking tests to measure functions including visual acuity and contrast sensitivity. This differs from current modes of trial data collection, whereby physicians gather data based on interactions with their patients. The company is hoping to increase the volume of data being collected by making it easier for patients to report the progression of their conditions. This is just the latest instance of the pharmaceutical giant using digital technology to bulk up its data collection efforts for the research and development of new treatments. Novartis is padding its offerings with digital services to differentiate itself from the competition and become more attractive to clients. The drugmaker is one of the most active investors in healthcare technologies, including telemedicine platforms, point-of-care diagnostics, and oncology therapeutics, according to CB Insights. In March, for instance, Novartis partnered with digital therapeutics firm Pear Therapeutics to improve the treatment of schizophrenia and multiple sclerosis, and it announced plans to work with Science 37 to develop “remote trial” technology that uses video conferencing to lower the barrier of entry for recruiting study participants.

NVIDIA JOINS THE RANKS OF TECH COMPANIES APPLYING AI TO MEDICAL IMAGING: The US tech company is developing an AI platform, dubbed “Clara,” to create a virtual medical imaging platform, according to MobiHealthNews. The service will be completely virtualized and “live” within a hospital’s data set. Radiologists and doctors will be able to use the platform to provide clinical decision support, such as by sorting through thousands of images including ultrasounds, X-rays, and CT scans to identify specific ailments. Importantly, the company is ensuring that Clara is compatible with a range of existing medical equipment. That means hospitals, clinics, and care facilities don’t need to buy new instruments to use the platform. It’s likely Nvidia chose to pursue medical imaging as it gears up for an eventual rollout to the broader medical industry. Computer vision — the form of visual AI used in medical imaging — is an established and tested field in several industries and has been actively explored within health by several companies, including Microsoft, Amazon, and several arms of Google-parent Alphabet. By establishing an early focus on medical imaging, the company can build up connections and clinical trials to serve as an entry-point to other segments of healthcare, such as genomics and drug discovery. Eventually, the company hopes to use AI to develop and deliver precision medicine.

IN OTHER NEWS:

• American Well and Philips have added a telemedicine feature into the Philips Avent uGrow app for parents of young children, according to MobiHealthNews. The new offering is the first to come from their partnership announced in January 2018.
• Proteus Digital Health, a digital therapeutics company, revealed its pipeline of 31 digital medicines for a range of illnesses including mental health, cardiovascular conditions, and oncology. Digital therapeutics are gaining the attention from larger pharma companies and health systems alike because of early signs they'll garner high retention rates. Patients using Proteus' DigiMeds take them around 90% of the time, compared with traditional medications, which sit at around 50%.  
• The US Drug Enforcement Agency is tapping Google Maps to help the Agency engage with consumers for the third National Prescription Drug Take Back Day, according to MobiHealthNews. The event is one of the US government’s efforts to tackle the opioid crisis by providing people with a way to safely dispose of expired and unnecessary prescription drugs. The DEA is working with Google to make it easier for consumers to locate nearby drop-off locations. Users can enter their zip code to find the nearest site or scroll over a map of their location to find a center. 

Join the conversation about this story »

NOW WATCH: Inside Cook Out, the South's most underrated restaurant