Well I, for one, have gained a great deal of effective psychological advice from ChatGPT - advice that has significantly altered my internal state and is changing my life for the better. My friend (an experienced software developer) uses AI to review code and look up obscure syntax. According to him, it functions at the level of a “very well-read intern”. A scientific collaborator I used to write data-analysis scripts for almost never asks me for help anymore. She is very happy because Gemini is good enough to do almost all of what she needs. A digital artist I know often edits promising AI-generated pictures rather than drawing from scratch. A very no-nonsense coworker increased his productivity so much using Copilot that he voluntarily arranged to give a talk to the rest of the developers about how great it was.
All these are people I know personally. I don’t think an AI can write a B- paper (it can write something that looks like a solid paper until, as you say, the sources are checked) but it’s already helping many professionals work more efficiently. For now, this looks like a warm-and-fuzzy “AI works together with humans rather than taking their jobs” scenario but every one of the people I’ve mentioned thinks that the technological development that lets AI take his or her job could happen any day now.
I’m not saying that the current boom in AI investment isn’t a bubble, but as Noah Smith points out, the railroad and dot-com bubbles that it’s being compared to were about technologies that really did change the world not long after the bubbles popped. The bubble investors weren’t as much wrong as they were just too early.
Depends on what you mean by “drastically”. The guy evangelizing to the rest of the company was claiming an increase in productivity of about 30%. I’d say that’s a big increase, but I wouldn’t use the word “drastic”.
Pumping out a B- paper means being able to replace the average college graduate.
So where is that happening? Because we’ve already got the B- papers, assuming no sources are checked.
Well I, for one, have gained a great deal of effective psychological advice from ChatGPT - advice that has significantly altered my internal state and is changing my life for the better. My friend (an experienced software developer) uses AI to review code and look up obscure syntax. According to him, it functions at the level of a “very well-read intern”. A scientific collaborator I used to write data-analysis scripts for almost never asks me for help anymore. She is very happy because Gemini is good enough to do almost all of what she needs. A digital artist I know often edits promising AI-generated pictures rather than drawing from scratch. A very no-nonsense coworker increased his productivity so much using Copilot that he voluntarily arranged to give a talk to the rest of the developers about how great it was.
All these are people I know personally. I don’t think an AI can write a B- paper (it can write something that looks like a solid paper until, as you say, the sources are checked) but it’s already helping many professionals work more efficiently. For now, this looks like a warm-and-fuzzy “AI works together with humans rather than taking their jobs” scenario but every one of the people I’ve mentioned thinks that the technological development that lets AI take his or her job could happen any day now.
I’m not saying that the current boom in AI investment isn’t a bubble, but as Noah Smith points out, the railroad and dot-com bubbles that it’s being compared to were about technologies that really did change the world not long after the bubbles popped. The bubble investors weren’t as much wrong as they were just too early.
Seek professional help. Don’t use llms as your phycologist.
Any “experienced” developer that says these AI tools have drastically increased their productivity is full of shit
Depends on what you mean by “drastically”. The guy evangelizing to the rest of the company was claiming an increase in productivity of about 30%. I’d say that’s a big increase, but I wouldn’t use the word “drastic”.
A lose of 30% sure.