TOI want to say too many (different) things to too many people. We need better ways to talk – and think – about it. Sign, Drew Breuniga talented cultural anthropologist and geek, who has come up with a clear categorization of technology into three use cases: gods, interns and gears.
“Gods,” in this sense, would be “superintelligent artificial entities that do things autonomously.” In other words, the AGI (artificial general intelligence) that OpenAI’s Sam Altman and his group are trying to build (at inordinate cost), while warning that it could be an existential threat to humanity. The gods of AI are, Breunig says, “human replacement use cases.” They require gigantic models and stupendous amounts of “computation”, water and electricity (not to mention the associated CO).2 emissions).
“Interns” are “supervised co-pilots who collaborate with experts, focusing on hard work.” In other words, things like ChatGPT, Claude, Llama, and similar large language models (LLMs). Their defining quality is that they are intended to be used and supervised by experts. They have a high tolerance for errors because the experts they help verify their production, preventing embarrassing mistakes from moving forward. They do the boring work: remembering documentation and navigating references, filling in the details after defining the broad strokes, helping with idea generation by acting as a dynamic sounding board, and much more.
Finally, “gears” are humble machines that are optimized to perform a single task extremely well, usually as part of a pipeline or interface.
Interns are mainly what we have now; They represent AI as a technology that augments human capabilities and are already widely used in many industries and occupations. In that sense, they are the first generation of quasi-intelligent machines with which humans have had close cognitive interactions in work environments, and we are starting to learn interesting things about how well those human-machine associations work.
One area where there are outlandish hopes for AI is healthcare. And rightly so. In 2018, for example, a collaboration between AI researchers at DeepMind and Moorfields Eye Hospital in London significantly accelerated the analysis of retinal scans to detect the symptoms of patients who needed urgent treatment. But in some ways, while technically difficult, that was a no-brainer: machines can “read” scans incredibly quickly and select those that need specialized diagnosis and treatment.
But what about the diagnostic process itself? Sign an intriguing American study published in October in the Journal of the American Medical Associationwhich reported on a randomized clinical trial on whether ChatGPT could improve the diagnostic capabilities of 50 practicing physicians. The boring conclusion was that “the availability of an LLM to physicians as a diagnostic aid did not significantly improve clinical reasoning compared to conventional resources.” But there was a surprising trick: ChatGPT alone demonstrated higher performance than both groups of doctors (those with and without access to the machine).
EITHER, like him New York Times summed it up“Doctors who received ChatGPT-4 along with conventional resources fared only slightly better than doctors who did not have access to the bot. And, to the researchers’ surprise, ChatGPT alone outperformed the doctors.”
More interesting, however, were two other revelations: the experiment demonstrated doctors’ sometimes unwavering belief in a diagnosis they had made, even when ChatGPT suggested a better one; and also suggested that at least some of the doctors didn’t really know how best to exploit the tool’s capabilities. Which in turn revealed what AI stands for, such as Ethan Mollick I’ve been saying for eons: that effective “rapid engineering” (knowing what to ask of an LLM to get the most out of it) is a subtle and little-understood art.
Equally interesting is the effect that collaborating with an AI has on the humans involved in the partnership. At MIT, a researcher performed an experiment to see how well materials scientists could do their jobs if they could use AI in their research.
The answer was that AI assistance really seems to work, as measured by the discovery of 44% more materials and a 39% increase in patent applications. This was achieved because AI performed more than half of the “idea generation” tasks, leaving researchers the task of evaluating candidate materials produced by models. So the AI did most of the “thinking,” while they were relegated to the more mundane task of evaluating the practical viability of ideas. And the result: the researchers experienced a sharp reduction in job satisfaction!
after newsletter promotion
Interesting, n’est-ce pas? These investigators are high-status people, not low-status agents. But suddenly, collaborating with an intelligent machine made them feel like… well, cogs. And the moral? Be careful what you wish for.
what i have been reading
camera piece
What if echo chambers work? is a striking essay that highlights a liberal dilemma in the era of Donald Trump.
Savings plan
A sharp analysis by Reuters is Charting the path for Elon Musk’s efficiency campaign.
inventive thinking
The fabulous and wise essay by Steven Sinofsky On the cost of being a disruptor It’s about innovation and change.