GPT/LLM links

Ethan Mollick's overview; The Zvi on the supply of fakes; Parents using ChatGPT to help their relationships with their children

Ethan Mollick writes,

the actual implications of what this phase of AI will mean for work and education is currently unknowable. It is unknowable to all of us who don’t have insight into what the AI labs have planned, but it is also actually unknowable to them. I guarantee that the people at Google and OpenAI and Microsoft do not know the implications of AI for YOUR job or YOUR company or YOUR education, or even all the ways in which the systems they are building will ultimately be used, for good or bad.

Yes, these are early days. Again, it reminds me of the state of the WWW in 1993, when many prognostications proved wrong. If there is one mistake that pervades the popular understanding of the new artificial intelligence, it is denying the possibility of use cases that involve creativity and empathy. We seem determined to hold onto creativity and empathy as qualities that make us “uniquely human.”

Zvi Mowshowitz writes,

As many point out, AI only matters [for affecting the prevalence of misinformation] when the quality [I think he means quantity] of supply matters. Which, in most circumstances, it mostly doesn’t.

Ethan Mollick: AI makes it easy for folks to make fake images & videos that are almost impossible to detect. I thought this would be a bigger deal in clouding the information environment (and it certainly will be). But people keep falling for video game footage & old pictures. AI is overkill.

Volokh Conspiracy is latest to point out that demand for political misinformation is a bigger danger than supply. AI increases supply, demand remains constant. If that demand is inelastic, as it seems to be, increasing supply will matter little.

I said this years ago about the spam problem. I thought that we should attack demand, not supply. Years ago, the most prevalent spam messages had the subject line “Enhance your penis!” I suggested that instead of punishing the sellers of penis enhancement products, we should punish the buyers. Get rid of demand, and you will get rid of supply.

How would you get rid of the demand for misinformation on Twitter? Off the top of my head, a demand-focused approach would be for Twitter to have dummy accounts that put out some unambiguous misinformation and find out who hearts it. If you heart misinformation, your account is suspended for a week.

For Axios, Erica Pandey and Megan Morrone write,

Foxx says he isn't trying to fool his daughter, just to connect. The last time he sent her a poem, he adds, she responded with "Thank you, Daddy" and a heart — "and then she was like, 'Did you use ChatGPT?'"

Pointer from Instapundit. This is another case of ChatGPT being used for its “right brain” (creative, empathetic) rather than its “left brain” (logical, analytical).

I would say that the father is not outsourcing his parenting to ChatGPT. He is using it as a partner, the way you might use your spouse or a sibling as a partner.

substacks referenced above:

@

AI #33: Cool New Interpretability Paper

@

The shape of the shadow of The Thing