For me it is powerful now, I work (remote) 8 it jobs simultaneously, earnings have increased 8-fold. I use generative deepfakes and conversational AI for simultaneous meetings.
I simply use Phind to generate functions or remind me how to use something, nothing fancy. I've got some basic prototypes for trading models but that's far from production (not really "useful" currently).
Generate some code, basic stuff as it fails on complicated functions using specific libraries. Keep in mind the LLMs mostly copy/combine already existing things, so if I want something unique, it will likely fail. It still saves me time and makes my life easier. People around me use it as well, in legal profession. They might need a simple letter and they only give general keywords for the LLM or ask LLM to fix typos or malformed sentences. For them saving the 15 minutes is beneficial.
AI still struggles with certain things, but damn, a Pulp Fiction Terminator mash up would make for a very interesting movie.
I am with @d08, I no longer write my own VBA subroutines. chatGPT and Gemini do a better job and infinitely faster with almost no bugs.
https://chatgpt.com/ https://gemini.google.com/app https://www.perplexity.ai/ https://www.meta.ai/ https://claude.ai/
LLMs are obviously not useless but this is the first month I can clearly see in my chatGPT chats that I did not use it enough to justify paying for it. I think like all bubbles, we have massively underestimated how huge this bubble has grown to while inside it. I remember waiting for Chinese food during the housing bubble in 2007 or 2008 and reading a Newsweek story about a 30 something couple from Portland that had retired to flip houses. It was so obviously ridiculous and not going to work long term but look at us now. We are so far beyond that in terms of ridiculous expectations that it is not even comparable. The fundamental error is not that LLMs are useless but that the market has priced forward an artificial mouth as if it is an artificial brain.
I found this interesting article from Goldman https://www.goldmansachs.com/intell...-too-much-spend-too-little-benefit/report.pdf I also read the paper from Daron Acemoglu this weekend. Acemoglu is in the Goldman article. https://economics.mit.edu/sites/default/files/2024-05/The Simple Macroeconomics of AI.pdf Acemoglu paper is 54 pages of pure speculation though, not really worth the time although I guess I largely already believed what he is arguing in the paper. The Goldman article raises many great points. It also concludes that the AI trade has room to run because bubbles take a long time to pop. Everyone already knows this though so I imagine we are closer to the end than we think.
agreed. I use it to generate some code, more so I know what to pass to various libraries. It's better than stackoverflow. Also, used it to create agreements (not sure if they are legally binding). AI will make productive people more productive. Just like the internet did. But you still need to know the foundations of what you're doing which is what most people never do