U.S. supremacy in A.I. may hinge on these proposed policies
Nearly a year ago, the Biden Administration debuted the National Artificial Intelligence Research Resource Task Force, intended to help the U.S. fully capitalize on progress in A.I.
The group includes a mix of government officials like U.S. deputy chief technology officer Lynne Parker, academics such as Stanford University’s Fei-Fei Li, and industry experts like Andrew Moore, director of Google Cloud’s A.I. unit.
On Friday, the task force issued recommendations to lawmakers about how to ensure that U.S. businesses and universities have the necessary resources to excel in machine learning. Their recipe: Making sure everyone in the U.S. has access to enough data and computing power to compete against rivals like China.
The task force’s recommendations are intended to help “lower the barriers” for companies to adopt A.I. technologies while also ensuring that they do so in a way that protects people’s privacy and doesn’t weaken cybersecurity, according to a summary of the report that the task force shared with Fortune.
Daniela Braga, a task force member and CEO of the enterprise startup Defined.ai (formerly DefinedCrowd), hopes the recommendations lead to the government creating a kind of A.I. marketplace, in which companies and research groups can pay to access datasets and machine-learning tools.
“It’s really like a marketplace with all the data at the U.S. government scale,” Braga said. “You will have your data from healthcare, from energy, from climate.”
Ultimately, the project is supposed to even the A.I. playing field so that not only Big Tech—Google parent Alphabet, Amazon, and [hotlink]Apple, for example—have access to the mountains of data and major computing resources necessary to make A.I. bloom.
Braga, who once worked at Microsoft, acknowledged the contributions Big Tech has made to A.I., particularly regarding deep learning and neural networks, the software that can discover and act on patterns it finds in enormous quantities of data. Online ad companies like Alphabet and Meta, in particular, have been able to obtain enough data over the years to greatly enhance their A.I. capabilities.
Today, however, privacy laws have made it more difficult to amass as much data, which has helped to solidify Big Tech’s dominance, Braga explained. The A.I. marketplace project is also intended to help organizations more easily keep up with China’s A.I. industry, which is “the big competitor here,” Braga said.
Indeed, Meta CEO Mark Zuckerberg has increasingly sounded alarms about the threat of Chinese companies like TikTok owner ByteDance to U.S. businesses.
Braga acknowledged the irony of the A.I. task force pointing to both U.S. tech giants and the Chinese tech industry as creating potential barriers to smaller companies and research groups taking full advantage of A.I. As Fortune previously reported, the U.S. is considering regulating Big Tech, which some experts speculate could have the unintended consequence of slowing A.I.’s progress in the U.S. while letting China leap ahead.
“I still see both sides,” Braga said about Big Tech’s contributions to A.I. and the argument that the U.S. needs these tech behemoths to dominate A.I. because of rising competition from China.
“But man, they really eat our lunch today,” Braga said. “It’s really difficult to compete.”
As for how the marketplace would work, Braga said companies and academic institutions would be able add their data to the marketplace. Companies would have to pay to access the marketplace, and if they contribute data or tools, they could get discounts, which could entice participation. While it’s unclear which government agency would oversee the project, it needs a sustainable business model, she said.
“It cannot be just funded by the government,” Braga said.
Fortune would love to hear from Eye on A.I. readers about ways the U.S. government can spur A.I. innovation in the private sector. Send my colleague Jeremy Kahn your thoughts.
On a personal note, this is my final edition of the Eye on A.I. newsletter, which my extremely capable and amazing colleague Jeremy will be taking over. It’s been an incredible past two years writing about A.I. and business for all you awesome readers. You all let me know when I failed to understand some sort of complicated nuance about A.I. and cheered me on when I (finally) had something unique to say. Like a neural network, I learned from all of you and your input. I’ll share more about my next phase sometime soon. Keep on reading!
A.I. IN THE NEWS
Delete your face. The United Kingdom’s Information Commissioner’s Office (ICO) has ordered facial-recognition startup Clearview AI to delete all photos of U.K. residents that the company has amassed over the years. Additionally, Clearview AI must pay the ICO a fine worth over $9 million U.S. dollars. From the ICO: Clearview AI Inc has collected more than 20 billion images of people’s faces and data from publicly available information on the internet and social media platforms all over the world to create an online database. People were not informed that their images were being collected or used in this way.
Waymo comes to downtown Phoenix. Waymo is expanding its self-driving taxi operations to downtown Phoenix, TechCrunch reported, bolstering the Alphabet subsidiary’s presence in the city, where it has been testing its driverless cars since 2016. In March, Waymo said it would "soft launch" its autonomous ride-hailing service for its employees in San Francisco.
Microsoft gives a big hug to A.I. Microsoft has partnered with the A.I. startup Hugging Face that will make the company’s A.I. tools available on the Azure cloud computing service. Hugging Face, which recently raised $100 million in a series C funding round, has become popular with A.I. researchers and companies who use the startup’s online repository of A.I.-related tools, including data and machine learning models.
A.I. hits Vermont. Norwich University has received a $4 million federal grant to establish a new research lab focusing on machine learning and quantum computing. “ I am confident that Norwich University's new center will lead the nation in developing the leaders to make the most of this technological revolution for the benefit of people everywhere,” U.S. Sen. Patrick Leahy, D-Vt., said in a statement.
EYE ON A.I. TALENT
Alphabet’s DeepMind A.I. subsidiary has hired Ian Goodfellow as an “individual contributor,” according to a Bloomberg News report that cited unnamed sources. Goodfellow was recently Apple’s director of machine learning, but left the tech giant after reportedly contesting its return-to-work policies. Goodfellow is a highly respected A.I. researcher who developed the so-called generative adversarial network technology that’s used to create deepfakes, which are computer-generated videos, photos, and audio that appear to be real. In 2019, Goodfellow was selected to be a member of Fortune’s 40 under 40 list.
TripleBlind chose Craig Gentry to be the enterprise startup’s chief technology officer. Gentry was previously a research fellow at the Algorand Foundation and a research scientist at IBM.
EYE ON A.I. RESEARCH
When A.I. and business fail. Researchers from the University of Oxford and the University of Strathclyde submitted a paper to the International Journal of Information Management Data Insights detailing the inflated expectations some business leaders may have regarding deep learning. Although deep learning has become a powerful technology that’s helped improve the ability of computers to recognize images in photos, among other things, it has yet to overtake more traditional statistical computing techniques in several business-focused tasks.
For instance, simpler statistical methods and more basic machine learning appear to work just as well as deep learning on several business problems that rely on so-called structured data, or information stored in rows and columns. Some of the business cases the authors present include predicting whether insurance policy holders will file claims and predicting whether certain marketing or sales campaigns lead to sales.
I’ve previously covered deep learning’s challenges working as well as other statistical methods in a Fortune feature in which I examined how companies like Genentech and Goldman Sachs were attempting to create neural networks that can analyze structured data. The goal was to create deep learning techniques that surpassed traditional methods, thus leading to potential breakthroughs in personalized medicine or financial investing.
FORTUNE ON A.I.
Tesla relegated to the backseat in Cathie Wood’s ARK fund as woes mount for Elon Musk—By Christiaan Hetzner
What’s up in A.I. The analyst firm CB Insights published its latest “State of AI” report covering financial activity during the first quarter of 2022. Here’s some of the takeaways from the report:
- Total A.I. funding in the first quarter of 2022 was $15.1 billion, representing a 12% drop from the fourth quarter of 2021.
- Insight Partners funded 16 A.I. startups, topping other venture capital firms in A.I. investments. Tiger Global came in second with 15 A.I. startup investments, followed by GV (formerly Google Ventures), which funded 12 A.I. startups.
- Securonix, a cybersecurity startup that uses machine learning, raised a $1 billion funding round, representing the largest A.I.-related deal in the most recent quarter.
- Investment in Chinese A.I. companies dropped 53% quarter-over-quarter to $1.6 billion in the most recent quarter, in line with the country’s overall quarterly decline of venture capital funding.
Our mission to make business better is fueled by readers like you. To enjoy unlimited access to our journalism, subscribe today.