Weekly AI recap: NYT sues OpenAI, Bezos invests in online search startup

“Defendants’ generative artificial intelligence (‘GenAI’) tools rely on large-language models (‘LLMs’) that were built by copying and using millions of The Times’s copyrighted news articles, in-depth investigations, opinion pieces, reviews, how-to guides and more,” the lawsuit states. The Times’ lawyers also argued that, though OpenAI scraped content “from many sources, they gave Times content particular emphasis when building their LLMs – revealing a preference that recognizes the value of those works.”

This is the first lawsuit to be filed against a major news organization against OpenAI, the company at the forefront of the wave of generative AI technology that’s transformed the tech industry and the information landscape over the past year. It follows similar lawsuits that have been filed against the AI company by prominent authors, including Jonathan Franzen and George R.R. Martin, who claim that their works have been illegally used to train a large language model. Given the novelty of generative AI, the courts are just beginning to grapple with the complicated legal questions that are raised by the technology – particularly those around copyright law.

The new lawsuit claims that the scraping of content from the Times has enabled OpenAI and Microsoft to produce content – through Microsoft’s GPT-4-equipped Bing search engine, for example – which users might consult as an alternative to the Times, thereby potentially having a negative financial impact upon the latter by cutting subscription and ad revenue. It also makes the case that the tendency of LLMs to “hallucinate,” or fabricate information, poses a substantial reputational risk to the Times; the lawyers cite an instance in which Bing Chat, asked to replicate a paragraph from a Times article, created information – including quotes – “that appear nowhere in The Times article in question or anywhere else on the internet.”

While it does not specify a monetary demand, the lawsuit argues that the defendants should be held “responsible for the billions of dollars in statutory and actual damages that they owe for the unlawful copying and use of The Times’s uniquely valuable works.” It also requests the “destruction” of all LLMs and training data owned by OpenAI or Microsoft which have incorporated copyrighted materials from the Times.

Jeff Bezos invests in up-and-coming AI-powered search start-up

The Wall Street Journal reported yesterday that Amazon founder and chair Jeff Bezos, along with a cohort of venture capitalists, has invested a cumulative $74mn in Perplexity AI, a start-up company based out of a coworking space in San Francisco that's currently staffed by fewer than 40 employees. The new injection of capital gives the company a valuation of $520mn, according to the Journal.

Founded in 2022, Perplexity AI leverages its own AI technology, along with OpenAI’s, to provide summaries to user search queries. The company is banking on the idea that users will eventually abandon the current model of online search, with its long list of links to other websites, in favor of a more direct and immediate response. “If you can directly answer somebody’s question, nobody needs those 10 blue links,” Avarind Srinivas, Perplexity AI’s CEO and a former researcher at OpenAI, told the Journal.

Though Google and Microsoft – the two current leaders of the online search industry – have both begun to integrate generative AI into their search engines, links remain the primary mode of search query response in either case.

Perplexity AI could potentially incorporate ads in the future as a means of generating revenue, according to the Journal. Yesterday’s report also noted that former Google executives Jeff Dean and Susan Wojcicki have both previously made personal investments in the start-up.

Chief Justice Roberts looks ahead to future impact of AI on legal field

In his annual Year-End Report to the Federal Judiciary, US Chief Justice John Roberts, Jr. highlighted the growing role of AI within the legal profession, and called attention to some salient risks and benefits which the technology presents to judges. “As 2023 draws to a close with breathless predictions about the future of artificial intelligence, some may wonder whether judges are about to become obsolete,” Roberts wrote in his report, which was dated December 31. “I am sure we are not – but equally confident that technological changes will continue to transform our work.”

Roberts predicts, on the one hand, that AI could democratize access to legal resources and information. “For those who cannot afford a lawyer, AI can help,” he wrote. But he also underscored his belief in the irreplaceable role of human intuition and judgement in the courtroom: “Nuance matters: Much can turn on a shaking hand, a quivering voice, a change of inflection, a bead of sweat, a moment’s hesitation, a fleeting break in eye contact. And most people still trust humans more than machines to perceive and draw the right inferences from these clues.”

Citing a long history of technological aversion within the legal profession – he notes that the Supreme Court did not house a photocopier until 1969, nearly thirty years after the machine was invented – Roberts encourages his colleagues to approach AI with “caution and humility,” but also with the knowledge that its widespread adoption within the field is more or less inevitable.

OpenAI announces launch of GPT Store

In an email sent out yesterday to verified GPT builders – subscribers to GPT Plus and Enterprise who use GPT-4 to build their own AI “agents” – OpenAI announced that it will be launching its GPT Store next week.

In the Store, builders will be able to buy and sell custom agents, the most popular of which will be featured on “leaderboards,” according to a description on OpenAI’s website. “We will also spotlight the most useful and delightful GPTs we come across in categories like productivity, education and ‘just for fun,’” the company wrote. “In the coming months, you’ll also be able to earn money based on how many people are using your GPT.”

OpenAI first announced that users could begin creating custom GPT agents at DevDay 2023, its inaugural developer’s conference, in November. The GPT Store was originally supposed to become available to the public that same month, but its launch had to be postponed (twice) when the company almost imploded after its board fired and then subsequently rehired its CEO, Sam Altman.

Microsoft announces new AI key

Yesterday, Microsoft unveiled a new key – to be placed between the Alt and left arrow keys at the bottom-right of keyboards on new Windows 11 PCs – which will link users directly with Copilot, the tech giant’s proprietary AI feature.

Microsoft is the primary investor in OpenAI, the AI company whose meteoric rise sparked after it released ChatGPT to the public in late 2022. Last year, Microsoft began integrating OpenAI’s large language model (LLM), GPT-4, into programs like Microsoft 365 and Bing, promising customers a more dynamic and personalized experience. Google – which dominates the search industry but has been swimming in Microsoft’s wake in this burgeoning era of LLM-powered personal computers – followed suit shortly thereafter by releasing Bard, its own AI assistant.

In a blog post, Microsoft executive vice-president Yusuf Mehdi described the new Copilot key as “the first significant change to the Windows PC keyboard in nearly three decades [since the company introduced the Windows key]. We see this as another transformative moment in our journey with Windows where Copilot will be the entry point into the world of AI on the PC,” he wrote.

Mehdi added in his blog post that Microsoft will have its new Copilot key-equipped keyboard on display at next week’s Consumer Electronics Show (CES) in Las Vegas, “with availability beginning in late February through Spring, including on upcoming Surface devices.”

For more on the latest happenings in AI, web3 and other cutting-edge technologies, sign up for The Emerging Tech Briefing newsletter.