ChatGPT is about to revolutionize the economy. We need to decide what that looks like.
When Anton Korinek, an economist at the University of Virginia and a fellow at the Brookings Institution, reached out to the new generation of big language models like ChatGPT, he did what many of us do: He started playing with them to see if they worked. How can you help in his work? They carefully documented their performance in a paper in February, noting how well they handled 25 “use cases,” from brainstorming and editing text (very useful) to coding (very good with some help) math. to do (not great).
ChatGPT misrepresented one of the most basic principles in economics, says Korinek: “It got really bad.” But the mistake, easily noticed, was quickly forgiven in the light of the benefits. “I can tell you it makes me more productive as a cognitive worker,” he says. “Hands down, there is no question to me that I am more productive when I use a language model.”
When GPT-4 came out, he tested its performance on the same 25 questions he documented in February, and it performed far better. There were fewer examples of goods being made; It also performed better on math tasks, says Korinek.
Korinek says, as ChatGPT and other AI bots automate cognitive work, as opposed to physical work that requires investment in equipment and infrastructure, economic productivity can increase exponentially. “I think we could see a greater increase in productivity by the end of the year – certainly by 2024,” he says.
Who will control the future of this amazing technology?
What’s more, he says, in the long term, the way AI models can make researchers like himself more productive has the potential to drive technological progress.
That potential of large language models is already turning up in research in the physical sciences. Bernd Smit, who runs the chemical engineering lab at EPFL in Lausanne, Switzerland, is an expert in using machine learning to discover new materials. Last year, after one of his graduate students, Kevin Mike Jablonka, showed some interesting results using GPT-3, Smit asked him to demonstrate that GPT-3 was, in fact, a more sophisticated machine-learning tool than his group. Teaching is useless for studies. To predict the properties of compounds.
“He completely failed,” jokes Smit.
It turns out that after being fine-tuned for a few minutes with a few relevant examples, the model has been developed along with advanced machine-learning tools, especially for chemistry, to predict the solubility of a compound or its solubility. Answers basic questions about things like reactivity. Just give it the name of a compound, and it can predict different properties depending on the structure.