It’s been more than 18 months since ChatGPT’s debut sparked a rally that propelled AI chipmaker Nvidia to a $3 trillion market cap in June, but billionaire investor Ken Griffin isn’t buying into the hype around the technology — at least not yet.
Speaking to Citadel’s new class of interns in New York, the founder and CEO acknowledged that artificial intelligence has reached an “inflection point” with the deployment of large language models, but disputed that LLM-based tools like OpenAI’s ChatGPT will become more valuable in the next three years than the top talent it recruits directly from universities.
“For a number of reasons, I’m not convinced that these models will achieve that type of breakthrough in the near future,” Griffin said Monday, according to comments reported by CNBC.
As evidence, the Citadel boss cited self-driving cars, which have difficulty coping with so-called edge cases, which exceed the limits of statistically common traffic events.
Most autonomous vehicles are trained by feeding them images labeled with data describing the environment around them: for example, a computer can recognize whether what is crossing the street is a small child or simply a rubber ball. But it often happens that an autonomous car is faced with a situation it has never seen before and behaves incorrectly because it lacks human intuition.
“Machine learning models work much better when there is consistency,” Griffin said, justifying his skepticism.
The Citadel founder has tended to view AI as a tool to enhance human productivity rather than replace it. Last month, he told attendees at the Milken Institute Global Conference that while AI has driven huge investments in data center infrastructure and semiconductor chips like Nvidia’s for AI training, its impact on businesses remains limited.
That’s partly because executives haven’t figured out how to most effectively harness the power of AI. They just know they need to tell investors like Griffin that they’re doing it.
“I ask the question, ‘So tell me how you’re using AI in your business today,’ and you get big smiles and very enthusiastic responses,” he said, “that have almost nothing to do with AI.”
“PhD-level intelligence”
While having top-tier human talent has helped Citadel run one of Wall Street’s most successful hedge funds to date, OpenAI CTO Mira Murati believes her team will have a say in whether knowledge economy workers like Griffin’s Citadel staff can still outperform its GenAI tools in the future.
Speaking two weeks ago to students at his alma mater, Dartmouth College, Murati said that OpenAI’s LLM, a type of neural network known as a generative pre-trained transformer, has matured rapidly in the space of a few years.
While GPT-3 was, according to him, still comparable to a toddler in terms of intelligence when it debuted in 2020, the current GPT-4 released in March 2023 is already at an advanced high school level.
“In the next few years, we’ll be looking at PhD-level intelligence for specific tasks,” she said.
Given that GPT-4 has already passed the qualifying exams for many professions, including the bar exam, one might think that Murati is downplaying the power of his technology. It remains to be seen whether Griffin will change his mind about the value and cost of human workers when Murati deploys GPT-5.
