Join top executives in San Francisco on July 11-12, to hear how leaders are integrating and optimizing AI investments for success. Learn More
The steady march of eye-popping investments into companies developing large language models (LLMs) continues. The New York Times reported today that OpenAI rival Character AI has raised a fresh $150 million in a recent funding round led by Andreessen Horowitz that values the company at $1 billion — which adds it to the 2023 unicorn club, even though the company has no revenue.
The Silicon Valley-based Character AI was founded in 2021 by two former Google researchers: Daniel De Freitas, who previously led LaMDA at Google Brain, and Noam Shazeer, one of the researchers behind the Transformers architecture, the technology that underlies ChatGPT.
Character AI’s offering may seem, at first glance, to be as light and fun as an Instagram filter. The company offers AI chatbots that allow users to chat and role-play with, well, anyone — living or dead, real or imagined. Think historical figures like Queen Elizabeth and William Shakespeare, for example, or fictional characters like Draco Malfoy (Character AI includes warnings like “Remember: Everything Characters say is made up!”)
>>Follow VentureBeat’s ongoing generative AI coverage<<
Join us in San Francisco on July 11-12, where top executives will share how they have integrated and optimized AI investments for success and avoided common pitfalls.
But others see Character AI as an example of a tiny group of private companies that, along with Big Tech giants like Microsoft and Google, are entrenching and consolidating AI power. According to a new article in the Financial Times, this means that “a handful of individuals and corporations now control much of the resources and knowledge in the sector — and will ultimately shape its impact on our collective future.”
A systematic shift as industry increasingly dominates AI research
The article points out that AI experts refer to this phenomenon as “industrial capture,” which was detailed in a recent paper by MIT researchers in the journal Science called “The growing influence of industry in AI research.”
According to the paper abstract, recent successes by companies like OpenAI and the massive funding flowing into companies like Character AI — as well as Anthropic, DeepMind, Adept and Cohere (the latter two are also founded by former Transformer coauthors — is emblematic of a “systematic shift as industry increasingly dominates the three key ingredients of modern AI research: computing power, large datasets, and highly skilled researchers. This domination of inputs is translating into AI research outcomes: Industry is becoming more influential in academic publications, cutting-edge models, and key benchmarks. And although these industry investments will benefit consumers, the accompanying research dominance should be a worry for policy-makers around the world because it means that public interest alternatives for important AI tools may become increasingly scarce.”
The MIT research found that almost 70% of AI Ph.D.s went to work in industry — rather than academia — in 2020, compared to only 21% in 2004. The Science paper author, Nur Ahmed, also found that companies’ share of the biggest AI models has gone from 11% in 2010 to 96% in 2021.
Concerns about everything from AI auditing to compute power
The New York Times article about Character AI also highlighted the risk of power consolidation among AI startups and Big Tech.
“One of the concerns I have is that it will be a winner-take-all or a winner-take-most market — that a few big players will really dominate,” said Erik Brynjolfsson, a Stanford University economics professor and a senior fellow at the school’s Institute for Human-Centered AI, in the piece.
That consolidated power is also about who has the access to the massive compute necessary to run these LLMs, the article continued: Mike Volpi, a general partner with Index Ventures who has invested in Cohere, “estimated that such companies required at least $500 million to spend on raw computer power.”
And the Financial Times article also pointed out that because GPT-4 and other industry-built LLMs are “black box” models that are not open to researchers to examine, researchers “cannot replicate the models built in corporate labs, and can therefore neither probe nor audit them for potential harms and biases very easily.”
It doesn’t seem like the flow of funding to these highly-sought-after AI startups will slow anytime soon — rumors about new Cohere funding, perhaps from Google or Nvidia, have been swirling for months. And just last week, Adept raised $350 million for generative AI trained to use every software tool and API.
The question is, what does this shift to “industrial capture” mean for the rest of the AI landscape? According to the Financial Times article, the ball is in the court of the policymakers, who cannot “turn a blind eye.”
VentureBeat's mission is to be a digital town square for technical decision-makers to gain knowledge about transformative enterprise technology and transact. Discover our Briefings.