601–610 of 1001 posts
Earlier quoted context omitted.
It’s trivial to address this. You ask an actual expert. I don’t treat any water cooler conversation as accurate. It’s for fun and socializing.
Asking an expert is only trivial if you have access to an expert to ask!
One fundamental challenge to me is that if each training run because more and more expensive, the time it takes it to learn what works/doesn't work widens. Half a billion dollars for training a model is already nuts, but if it takes 100 iterations to perfect it, you've cumulatively spent 50 billion dollars... Smaller models may actually be where rapid innovation continues simply because of tighter feedback loops. O3…
AGI is the Sisyphean task of our age. We’ll push this boulder up the mountain because we have to, even if it kills us.
Any technology may kill us, but we'll keep innovating as we ought to. What's your next point?
Earlier quoted context omitted.
I'm in the same boat as the person you're responding to. I really don't understand how to get anything helpful out of ChatGPT, or more than anything basic out of Claude. > I've found that if you treat it more like a colleague, it works wonderfully. This is what I've been trying to do. I don't use LLM code completion tools. I'll ask anything from how to do something "basicish" with html & css, and it'll always output…
My approach is a lot of writing out ideas and giving them to ChatGPT. ChatGPT sometimes nods along, sometimes offers bad or meaningless suggestions, sometimes offers good suggestions, sometimes points out (what should have been) obvious errors or mistakes. The process of writing stuff out is useful anyway and sometimes getting good feedback on it is even better. When coding I will often find myself in kind of a rever…
Earlier quoted context omitted.
Just today I got Claude to convert a company’s PDF protocol specification into an actual working python implementation of that protocol. It would have been uncreative drudge work for a human, but I would have absolutely paid a week of junior dev time for it. Instead I wrote it alongside AI and it took me barely more than an hour. The best part is, I’ve never written any (substantial) python code before.
It would seem you don't care too much about verifying its output or about its correctness. If you did, it wouldn't take you just an hour. I guess you'll let correctness be someone else's problem.
I developed the library interactively, one API call at a time, in a manner akin to pair programming. Code quality was significantly better than I'd expect from $2000 worth of a GOOD mid-tier programmer — the code was well written, well organised, and comprehensively annotated. The code wasn't perfect, but a majority of faults had a basis in the underlying documentation being wrong or ambiguous.
The $20/month for Cursor Pro literally justified its cost in less than 10 minutes.
Earlier quoted context omitted.
This is a good read for some examples https://arxiv.org/abs/2203.14465 > This technique, the "Self-Taught Reasoner" (STaR), relies on a simple loop: generate rationales to answer many questions, prompted with a few rationale examples; if the generated answers are wrong, try again to generate a rationale given the correct answer; fine-tune on all the rationales that ultimately yielded correct answers; repeat. We show…
One issue with that is that the model may learn to smuggle data. You as a human think that the plain reading of the words is what is doing the reasoning, but (part of) the processing is done by the exact comma placement and synonym choice etc. Data smuggling is a known phenomenon in similar tasks.
Earlier quoted context omitted.
> but being this confidently wrong (and "lying" when confronted with it) is troubling. It works in politics, marketing, and self-promotion. If you use the web as a training set, those categories dominate.
Maybe they also trained the model on Sam Altman. ;)
LLMs have got to the point where they have (more or less) encoded everything ever written by humanity. And encoded it in a gazillion dimensioned matrix so they can take a query and predict, given everything ever written, what the most likely set of words are triggered by that query - and then produce that as a sane parseable output.
I mean this is beyond awesome
But, this reminds me of Aristotle reasoning from his current set of knowledge - and ending up with Eagles having three testicles.
No matter how much reasoning we do from current knowledge, we need to test that against reality - that LLMs are great but the scientific method is greater.
Google is in trouble, yes.
But humanity needs to hold on to science and it’s precepts or we spin into a new dark ages
Earlier quoted context omitted.
It would seem you don't care too much about verifying its output or about its correctness. If you did, it wouldn't take you just an hour. I guess you'll let correctness be someone else's problem.
I don't know the OP here, but in my experience a junior dev at an average company would likely not do much more than the AI would. These aren't your grandfather's engineers, after all.
What I’ve learned is that for the most part that AI revolution is not going to be because of PHD-level LLMs. It will be because people are better equipped to use the high-schooler level LLMs to do their work more efficiently.
We have some knowledge graph experiments where LLMs continuously monitor user actions on Slack, GitHub etc and build up an expertise store. It learns about your work, your workflows and then you can RAG them.
In user testing, people most closely associated this experience to having someone just being able to read their minds and essentially auto-suggest their work outputs. Basically it’s like another team member.
Since these are just nodes in a knowledge graph, you can mix and match expertise bases that span several skills too. Eg: A Pm who understands the nuances of technical feasibility.
And it didn’t require user training or prompting LLMs.
So while GPT-5 may be delayed, I don’t think that’s stopping or slowing down a revolution in knowledge-worker productivity.