January 31, 2024

AI Reporting

A selection of my Artificial Intelligence journalism, including Generative AI, ChatGPT, and AI Ethics.

Plastic Surgeons Are Using A.I. to Determine Beauty, One Zero

Will algorithms destroy the cultural diversity of what qualifies as pretty? When I surveyed around 20 plastic surgery conferences that took place this year, most included a session about A.I. and a discussion of software programs that do everything from measure facial attractiveness to recommend procedures for doctors to pitch to prospective clients. Heather Levites, a plastic surgery resident at Duke University School of Medicine, says that using AI gives her unprecedented insight on how patients feel about different procedures. But some experts say that handing over assessments of beauty to an algorithm may not be a good idea. “There may be a marginalization of values of beauty in other cultures,” warned plastic surgeon Dr. Jungen Koimizu.

In Zurich, Endri Dibra, founded Arbrea Labs, an AI-powered augmented reality imaging tool that would have let his mother see — in real-time — what her post mastectomy breasts would look like. Why were his mom’s expectations and her reality so different, he asked. “We have refined algorithms and AI tech that didn’t exist before,” he says. “Patients have the right to be better informed of the outcomes,” he says. There were other visualization tools for surgeries, but he felt were simplistic; they didn’t account for changes in lighting or factor in skin color (Black skin is prone to keloid scarring, for example) or account for the full body proportions (many depict limbless torso’s).

The Secret Life of Selfies: How a Beauty Tech Startup Is Using AI to Match Faces With Products, The Information

Inside Taiwanese unicorn Perfect Corp.’s billion-dollar bet on biometric beauty tech. For years, Hannah Williams’ go-to selfie app has been YouCam Makeup. Before posting a self-portrait to Instagram, the 25-year-old cashier from Morgantown, W. Va., will run the shot through YouCam’s augmented reality filters to whiten her teeth, remove stray pimples and smooth her complexion. She’s gotten a lot of value out of the app: It makes her looked “snatched,” she said. But each of Williams’ selfies offers even more value to Perfect Corp., YouCam’s Taiwan-based parent company.

Backed by investors including Snap, Chanel, Goldman Sachs Asset Management and Alibaba Group Holding, Perfect Corp. offers consumers and brands sophisticated try-on tools that allow customers to virtually test beauty products from their own home. For as little as $399 a month for enterprise clients, Perfect Corp.’s artificial intelligence platform—trained on hundreds of millions of faces like Williams’—can analyze users’ skin quality, match the exact shade of their skin to a corresponding product and overlay accessories. And pretty soon, the program may even label users’ personality traits, a feature that presents more dystopian possibilities. Recently, Perfect Corp. launched an AI Personality Finder, which promises to read your facial features for clues about what kind of person you are—and, by extension, what kind of products you might buy.

Will Your Cheatin’ Heart Tell on You? As Americans Lose Trust in Each Other, They’re Turning to Tech to Detect Lies, The Information

From AI-enabled ocular motor detectors, to partner-tracking spyware, to mail-in infidelity tests, tech tools are filling the gaps in an age of distrust. A few months ago, inside a soundproof room attached to his office, Dennis Cates, a retired police officer and founder of Mid Valley Polygraph in Porterville, Calif., directed his client to answer an awkward question. “Have you ever been intimate with your husband’s brother?” “No,” the client answered, clicking false on the screen in front of her. Unfortunately, her pupil dilation, blink rate and involuntary eye movements, as captured by an infrared camera logging 60 data points per second, in EyeDetect, an artificial intelligence–powered lie detection platform told another story. “You’ve failed,” Cates informed the woman.

Cheating is and has always been a reality in relationships—in the U.S., approximately 23% of married men and 14% of married women have cheated on their partner, according to data reported in the Journal of Social Sciences. But what is new is how technology adjudicates such acts of betrayal. Alongside a raft of relationship-mending apps like Fix A Fight, Relish and Lasting, developers are trotting out a number of new products as infidelity sniffers, some more ethically dubious than others.  EyeDetect tests are the core product of Converus, a Utah-based company that has raised $11.67 million from investors including Mark Cuban, Album VC, and Alta Ventures. EyeDetect’s initial focus was to curb corruption in Latin America, said Converus CEO Todd Mickelsen, a soft-spoken white man in his mid-50s, whose email sign-off is “truthfully.” Recent expansions include infidelity testing, employment screenings, and partnerships with law enforcement and immigration agencies, which use it to vet refugees and monitor registered sex offenders.

Know Thy Selfie: A Journey Into the Uncanny Valley of AI-Generated Avatars, The Information

Seemingly overnight, Lensa, Avatar AI and other generative AI selfie apps blanketed the internet.

At 7 a.m. on August 22, 2022, Yaron Inger woke up to find his business irrevocably changed. The co-founder and chief technology officer of Lightricks, which develops Facetune and other popular media-editing apps, scrolled through his newsfeed, his eyes flicking from one image to another. While Inger slept, Stable Diffusion, an open-source artificial intelligence text-to-image generator, had been released and was already shaking up the internet. The quality of the images it was cranking out was breathtaking. “It was very clear, from the very beginning, that we’re on the verge of something very big—and that we have to be there,” said Inger. Lightricks, which had raised $130 million at a $1.8 billion valuation less than a year before, immediately got to work.

The Artificial Intelligence Ethics Committee, Forbes

As artificial intelligence becomes deeply ingrained in the creation of our products, concern about whose interests they serve has spawned a number of gatekeepers. Their goal: to keep human welfare paramount in the code. Enter the ethics committee — officially. “Data bias exists, so question people when they say their answer comes from data,” said Nick Loui,  the CEO and founder of CivicFeed, a platform that uses AI to to make government actions more transparent. “Ask them how it was compiled.”

 In academia, you can look to Harvard’s AI Initiative, Oxford University’s AI Code of Ethics project, and AI4ALL, an Oakland-based nonprofit that partners with universities like Stanford and Berkeley, to name a few. Then there’s the Partnership on AI nonprofit, founded in 2016 by a collection of tech companies including Apple, Facebook, Google, IBM, and Microsoft. Their task is to develop best practices, create inclusive networks and look at the social impact of AI.

Will AI end the Art of Lying? OZY

Scientists and entrepreneurs are developing ways to harness swarm intelligence to reap the benefits while keeping people in the equation, along with their creativity, intuition, judgment and morality. One such outfit is Unanimous A.I., a Silicon Valley–based company that meshes human swarms with complex algorithms. In a recent study, one of its swarms collectively made 46 percent fewer errors when identifying fake smiles than individual participants.“Humans are not very accurate at telling if someone is being honest or deceptive,” says CEO Dr. Louis Rosenberg, who finds swarm intelligence not only a better predictor of truth but also an effective way of solving problems. He references nature’s swarms, such as birds, bees and fish. When it comes to food, shelter and survival, they outperform individuals and collectively make decisions for the greater good. “If a swarm acted like Congress, it would die,” he says, emphasizing that survival depends on cohesion.

Looking to the future, critics visualize swarms and emotional A.I. used in a number of troubling contexts, including politics, health care and law enforcement. But Katja Grace, an A.I. researcher at the Future of Life Institute, believes those concerns are blown out of proportion. “I expect this eventually will have some social implications, but it does not herald especially humanlike intelligence in general,” Grace says. “Being able to recognize a scared demeanor is very different from understanding what fear is.”

Meet Laura Montoya: The Queer Latina Trying to Build Bias Free AI, OZY

AI is a booming industry today, but Laura Montoya, who launched her company in September 2016, is worried that it’s becoming too homogeneous. Her goal with Accel.AI is to counteract that problem by training groups that are underrepresented in AI and make the industry more diverse. “AI tech is a direct reflection of the people who are engineering it, so any bias by these individuals will be reflected in the products they create,” Montoya says — something she’s seen many times with “tech bros” in Silicon Valley. Looking for examples? In 2009, HP’s imaging software couldn’t recognize Asian faces, and Harvard’s Project Implicit discovered that people automatically assign positive or negative behavior to different skintones.

Carlos Uranga, a Silicon Valley investor and former director of Singularity University’s Innovation Lab, heard of Montoya — and became intrigued with her proposal. “There’s an AI space race taking place right now,” he says, but Montoya’s approach of combining disciplines to get there is something he, a self-professed generalist, found appealing. “AI is the glue between robotics, computer vision [and]bioengineering,” he says.

Meet Frida Polli, The Neuroscientist Using AI to Level the Recruiting Field, OZY

Polli’s original plan wasn’t to build a diversity tool but to make the assessment system less biased, which ended up being the same thing. “There are historical biases baked into the current assessment models against women and minorities,” Polli says. Ideally, AI can fix that, though she’s aware that some warn predictive AI could kill jobs; she counters that it can also highlight talent shortages and suggest new paths of employment.

But not everyone is sold on the benefit of psychometric tests, AI notwithstanding. Science writer Annie Murphy Paul calls tests “often invalid, unreliable and unfair,” and Roger Dooley, neuromarketing expert, is skeptical. “I’d like to see the data showing that the tests really are effective at predicting job performance,” he said via email.

The Hiring Hackers, OZY, USA Today

 Hiring is just the start, as retention and culture fit is a huge issue for employers. This is what Jeff Barson is focused on at HireVue Labs, where he develops machine learning programs for recruitment. He’s working with algorithms which predict the future behaviors of employees, divining if candidates are motivated, trustworthy, ethical and have management potential. Barson can confidently answer yes or no, due to the high-level access to companies data that he’s been granted.  “It’s just data but it has never been available before,” Barson says. “Organizations know who they hire, but not why.” Barson says this is possible due to high-level access that his business has been granted by companies that share data about employee performance and salary changes. By tracking rising stars — and failures — on video interviews, HireVue tries to predict which traits denote success at different companies, and pair them with people who have matching characteristics.

The AI Advantage, Stylus

At MIT Technology Review’s EmTech Digital conference in San Francisco, speakers explored how advances in artificial intelligence (AI) are reshaping the underlying structure of every industry, in areas ranging from education to retail to law. Spotlights included China’s growing domination of AI development, the emergence of cloud-based AI services and man-machine interfacing.

Advances in deep-learning language translation, combined with the decreasing cost of hardware, are enabling international travelers to understand different languages in real time. “There’s a huge demand for multilingual translation,” said scientist Hua Wu, who spearheaded the portable pocket translator developed by Baidu, China’s largest search engine. There’s also an uptick in the cloud-based artificial intelligence (AI) services that internet companies are offering, which give their users everything from vision to language processing. “Cloud AI is not sexy, but it’s important,” said Oren Etzioni, chief executive of the Allen Institute for Artificial Intelligence in the US. “Some say it will democratize AI.” The addition of AI to the solution stack will play a big role in the IT sector’s predicted 5% growth.

The Autonomous Evolution, Stylus

“We don’t want a universe where you have to choose between safety and making money,” said Lior Ron, co­founder of Otto – a US self­-driving truck start­up acquired by Uber in 2016. Using mounted cameras, radar and lidar sensors, Otto’s AI is sophisticated enough to steer, brake and accelerate on the roads. The tech was demonstrated via a partnership with Budweiser, in which a driverless Otto truck drove 120 miles across Colorado to deliver 51,744 cans of beer. However, small business owners should not get over­ excited, warned Peter Lee, corporate vice­ president of Microsoft Research. “We’re seeing a long gate between the conception of ideas and their broad deployment,” he said.

Given businesses’ desire to weave AI into services, developers are offering one ­stop solutions for companies without data science teams. New York­ based start­up Clarifai does the heavy lifting for companies looking to integrate visual search by ingesting their media files (images and video) and tagging them in real time. “Humans can do a little bit of work and apply it to massive amounts of data,” said chief executive Matthew Zeiler.