An insider’s guide to AI and exponential technologies
Azeem, as a subscriber, I would love your input on this question, either here or in an edition should you so choose.
I see people describe AI as offering these wonderful things. However, when I interact with it (outside of Perplexity.ai, which was a deeply brilliant recommendation, thank you), I usually find the models to be a bit ... well, if they were my intern or assistant (not that I'd ever have one), I'd have fired them long ago.
I don't seem to get the miraculous results I see relayed from other people. I am not sure if the fault lies with me, or if models have been "dumbed down" due to computing costs or social implications of everyone suddenly getting a genius to work for them, etc. (I am not a conspiracy nut but that one does strike me as plausible.)
I constantly find errors reported as truth, or ChatGPT forgetting that it has a Bing plugin, or finding that code it offers doesn't work because some detail was overlooked, or so on and so on. This doesn't seem to correspond with the reports people give of AI coding whole applications for them (such as with the "Galactic Center" app released this week).
I am not sure if I am anthropomorphizing it as a human being and thus expecting higher levels of performance that AI is not at yet, or if the fault is mine (when compared to others' reported skills), or if I am perhaps not engaging with it in a proper back-and-forth and am instead simply getting impatient or angered when error-ridden work is returned and a sheepish "I'm sorry" is returned by ChatGPT.
(As an aside, I am sure that this will be a whole interesting realm of social science that may crop up - perhaps even of use for our communications with aliens or other intelligent species - trying to navigate the differences between how humans interact and how others do, and helping navigate and bridge those differences.)
You have full permission to reprint this comment should you wish to include it a newsletter. (I hope the simple quest for what seems to cause a problem - including the willingness to assume it's me - will not cause people to attack me.)
Thank you, this is a great article! :-) Can I have a small question about the Sunday Chart? What exactly is the "Daily average anomaly"?