Bulletins

OpenAI’s GPT-3 language model can now be customized for specific apps

The research outfit will now let all its API users create customized versions of GPT-3 that are fine-tuned for specific applications, such as call centers.

A call center.

OpenAI, DeepMind and others say they are dedicated to producing “beneficial” and “safe” AI.

Photo: Kuni Takahashi/Bloomberg via Getty Images

AI that produces human-like language for chatbots, video games, semantic search and customer service is getting a lot of attention lately. The latest news comes out of OpenAI, the research company known for its GPT-3 machine-learning model used to automatically generate sophisticated text. The research outfit will now let all its API users create customized versions of GPT-3 that are fine-tuned for specific applications.


And, in keeping with the low-code trend making AI more accessible to people without computer engineering Ph.D.s, it works to train AI models on those application-specific datasets — say, a sampling of a few hundred product reviews — with the proverbial “single line of code.” The idea is to make this form of content-generating AI faster and cheaper because developers don’t have to create and train AI systems from the ground up, which can be a pricey and time-consuming endeavor.

GPT-3 is the third iteration of the GPT natural language model. It employs a form of machine learning called few-shot learning, which can be trained using a "few" examples of natural language sentences. Few-shot learning is also the foundation for tools Meta is testing in its constant battle to detect disinformation and other content that violates policies on Facebook and Instagram.

As OpenAI has advanced its language AI, others including Nvidia and China’s Inspur AI Research have pushed the limits of natural language processing. Meanwhile, Google’s DeepMind unveiled new language AI research last week including a study of its Gopher language model.

OpenAI, DeepMind and others say they are dedicated to producing “beneficial” and “safe” AI. However, making machine-learning models more and more accessible and easy-to-use poses serious risks of even more prolific dissemination of false, sensitive or discriminatory content and information.

DeepMind’s recent research looks into these risks by creating a way to classify them. OpenAI also helps to mitigate the negative impacts of text-generating AI in the wild by reviewing and actively monitoring applications built with GPT-3.

Correction: This story originally identified OpenAI as an open-source company.

Latest Bulletins

Robinhood shares plunged late Thursday, falling below $10 after the company reported disappointing results and a revenue outlook that well below what Wall Street was expecting.

Keep Reading Show less

A New York law taking effect May 7 requires employers to disclose electronic monitoring of their workers. This is similar to workplace monitoring laws already in effect in Connecticut since 1998 and Delaware since 2017. Federally, employers may spy on their employees as they perform work duties.

Keep Reading Show less

Proponents of the America Competes Act of 2022, introduced this week in the U.S. House of Representatives, said the legislation would make the U.S. supply chain stronger and help the country's technology industry compete with China. The crypto industry is worried the sector will be collateral damage.

Keep Reading Show less

Tesla has sued a Chinese car influencer with 14 million followers for defamation, the company’s Chinese office confirmed to several media outlets on Wednesday. It’s the latest example of Tesla striking back at popular allegations in China about its safety issues, particularly around brake malfunctions.

Keep Reading Show less

Google is increasing its leave benefits for parents, parents who give birth and those caring for seriously ill loved ones, the company announced Thursday.

Keep Reading Show less

SoftBank COO Marcelo Claure may step down as soon as today, CNBC reported.

Keep Reading Show less

In Tesla's fourth-quarter earnings call Thursday, Elon Musk said the company won’t produce any new models this year.

Tesla reported overall strong earnings with revenue growing 65% year-over-year in the quarter, and automotive revenue rising to $15.97 billion. Musk addressed a range of issues during the call, including the persistent chip shortage, supply chain concerns and Tesla’s autonomous vehicle push.

Keep Reading Show less

Regulators in China said today they will allow AMD to acquire Xilinx, the Wall Street Journal reports. It’s the last regulatory hurdle the company needed to cross to complete the largest deal in its history.

Keep Reading Show less

About 100 company leaders are pressing their colleagues to improve the workplace for people with disabilities and evaluate their inclusion of people with disabilities, Disability:IN announced Thursday.

The letter’s newest signatories of the “CEOs Are IN” campaign include the heads of Micron Technology, Tripadvisor and other tech and health care professionals. Accenture, Intel and Microsoft are among the first companies to back the letter when it was first released in 2019. These companies pledged to make disability inclusion a business priority by participating in an assessment of their disability inclusion and equality practices.

Disability:IN CEO Jill Houghton said focusing on diversity, equity and inclusion efforts will help create sustainable companies, especially in light of the Great Resignation. “By encouraging CEOs to make disability inclusion a business priority, these 101 signatories are joining IN to positively transform lives and create financial and social impact,” Houghton said in a release.

The signatories pledged to take part in the Disability Equality Index, which is administered by Disability:IN and the American Association of People with Disabilities. Participants of the assessment receive a score between 0 and 100, with scores over 80 indicating that they are the “Best Places to Work for Disability Inclusion,” according to the release. The DEI looks at criteria within six categories including culture and leadership, community engagement and supplier diversity. The organization added questions related to board diversity this year, and the findings from those questions will be released in July, according to Disability:IN's Houghton.

Of the 319 companies that took the examination last year, 272 scored 80 or above, Houghton said. Adobe, Dell and Cisco were among dozens of companies to receive a score of 100 in 2021; Amazon and VMware were given scores of 90%; and Snap, Micron and Workday received scores of 80%. Registration for the Disability Equality Index is currently open, and results are released in July.

The Diem Association, a Meta-backed crypto project, is winding down and selling its technology to Silvergate Capital for $200 million, the Wall Street Journal reported on Wednesday.

Keep Reading Show less

The Amazon Labor Union has reportedly gotten a "sufficient showing of interest" in holding a union election among JFK8 warehouse employees in Staten Island, New York.

Keep Reading Show less

Shares of LendingClub fell in after-hours trading Wednesday after posting weaker-than-expected revenue outlook, despite reporting strong results. The fall below $20 wiped out almost all the gains the company saw in a rally that ran from July to November as its neobank strategy showed strength, taking shares from around $16 to as high as $49. With inflation on the rise and other threats looming, that strategy's freshly in question.

Keep Reading Show less

A Walmart-backed fintech startup is buying two companies in an effort to become the latest super app to handle all of a consumer's financial needs.

Keep Reading Show less

UBS is buying robo-adviser Wealthfront for $1.4 billion, the financial services giant said Wednesday.

Keep Reading Show less

Amazon has stopped paying warehouse workers to sing the company's praises on social media, according to the Financial Times.

Keep Reading Show less

The Business Roundtable, whose members include more than 230 CEOs of some of the world’s largest companies, want “rational” and “flexible” government regulations for AI. The group, which includes CEOs of tech, financial services and defense industry giants such as 3M, Amazon, Bank of America, Google, Mastercard, Northrop Grumman, Oracle and Verizon, published a set of guidelines Wednesday for businesses implementing AI and recommendations for policymakers.

Keep Reading Show less

Google broke the internet last year with its plan to break the internet last year.

Of course, I’m talking about FLoC, Google’s poorly named plan to kill off the third-party cookie and allow advertisers to target “cohorts” of anonymous users instead. It managed to piss off everyone, with privacy groups calling it invasive and the ad industry calling it anticompetitive.

Keep Reading Show less

The Diem Association, a controversial Meta-backed cryptocurrency project, is reportedly looking to sell its assets to return capital to investors, Bloomberg reported.

Keep Reading Show less

Boris Johnson said on Tuesday that he is talking to the U.S. about a possible Russian ban from the SWIFT system, which undergirds global payments. Amid increasing threats of a Ukraine invasion, Western governments are looking for deterrence measures.

Keep Reading Show less

Russia’s Ministry of Finance is opposed to the Central Bank of Russia’s proposed crypto ban, calling for regulation instead of a complete ban.

Keep Reading Show less

Twitter was abuzz Tuesday over a long thread written on Monday by Bolt CEO Ryan Breslow, who claimed that Stripe and Y Combinator are “mob bosses,” using “every power move imaginable” to block competitors from becoming successful.

In other words: Silicon Valley is cutthroat. Who knew?

Keep Reading Show less

Ripple has bought back shares it issued after securing $200 million in funding in 2019, an unusual move which CEO Brad Garlinghouse said underlined the crypto company’s momentum despite its ongoing legal battle with the SEC.

Keep Reading Show less

Nvidia's bold $40 billion bid for chip designer Arm has been in trouble for months, after regulatory impediments in Europe, the U.K. and the U.S. have threatened to disintegrate the deal. It might be time for Plan B.

Keep Reading Show less

In the last few years, Google tried out a new project for ad targeting, FLoC, in an effort to get rid of cookies. The system was supposed to group people by interests for advertising purposes, but received significant pushback from experts who believed it would exacerbate discriminatory and predatory ad targeting.

Today, the plan is dead. In its place: Topics.

Keep Reading Show less

Raven Software, the Activision-owned studio responsible for the hugely successful Call of Duty: Warzone, told employees on Monday it would begin embedding its quality assurance testers within different departments at the company, according to a staff email obtained by Polygon.

Keep Reading Show less
Bulletins