I don’t think the system has that much sophistication.
I do think they can “weight” the training set and feed it endless variations of “approved content” to be regarded as correct, and maybe also feed it other content to be identified as “incorrect” and rebutted from the approved content.
Last time I checked, Grok was still ‘woke’ (meaning: repeating the most common opinions published around the internet) - that’s been a couple of months. I was shocked that it was permitted to be accessed by the public back then.
I’m beginning to think those people are already lost… anyone who is going to fall for this has already fallen for similar things and reflexively distrusts anything that doesn’t back up their world views.
I also think people who are opposed to this are very unlikely to change their minds based on something controlled by someone like that.
Finally, we’ve been at this hard for almost a decade, I’m pretty sure the undecideds are lying to the pollsters.
Took 'em long enough to get around to this result, I’m shocked that the boy genius would let it operate in “bash me because that’s what the majority of the world thinks” mode for so long.
Probably just to save money really.
That always helps. It also helps politically that M$ is based in a country that’s outraging the Danish people on a fairly regular basis…
same thing with 365 but we do a lot of traveling
Back in 1990-something, I got our office using Ami Pro - it was a vastly superior word processor to anything else available at the time. Then, a couple of years later, we started sharing documents back and forth with business partners via dial-up internet and that was the end of Ami Pro, all our partners used M$ and file format translation / import / export was nigh impossible in those days.
I did this in my personal usage 20 years ago. I even was demonstrating to colleagues at work in 2005 how Open Office was better at integrating large numbers of digital photos into documents than Word was.
I have been more successful with baby steps like: “Write a python 3 program that converts X to Y.” Tweak prompt until that’s working as desired, then: “make it work recursively through all subdirectories” - and again tweak with specifics like converting the files in place, etc. Always very specific, also - force it to fix its own bugs so you can move forward with a clean example as you add complexity. Complexity seems to cap out at a couple of pages of code, at which point “Ooops, something went wrong.”
He’s developing a toxic relationship with his AI agent. I don’t think it’s the best way to get what you want (demonstrating how to be abusive to the AI), but maybe it’s the only method he is capable of getting results with.
I frequently find myself prompting it: “now show me the whole program with all the errors corrected.” Sometimes I have to ask that two or three times, different ways, before it coughs up the next iteration ready to copy-paste-test. Most times when it gives errors I’ll just write "address: " and copy-paste the error message in - frequently the text of the AI response will apologize, less frequently it will actually fix the error.
The first half dozen times I tried AI for code, across the past year or so, it failed pretty much as you describe.
Finally, I hit on some things it can do. For me: keeping the instructions more general, not specifying certain libraries for instance, was the key to getting something that actually does something. Also, if it doesn’t show you the whole program, get it to show you the whole thing, and make it fix its own mistakes so you can build on working code with later requests.
There’s a certain amount of: “if this isn’t going to take over the world, I’m going to just take my money and put it in something that will” mentality out there. It’s not 100% of all investors, but it’s pervasive enough that the “potential world beaters” are seriously over-funded as compared to their more modest reliable inflation+10% YoY return alternatives.
I ask AI to write simple little programs. One time in three they actually compile without errors. To the credit of the AI, I can feed it the error and about half the time it will fix it. Then, when it compiles and runs without crashing, about one time in three it will actually do what I wanted. To the credit of AI, I can give it revised instructions and about half the time it can fix the program to work as intended.
So, yeah, a lot like interns.
That’s what happens when you have Intel inside ;o)
(Yes, yes, I know, it’s the whole binary based floating point thing, not just Intel, although my Atari 800 BASIC interpreter implemented floating point in BCD, so it didn’t have that issue.)
More like 0.10 + 0.05 = 0.20, in this case.
It has also led to me saying “I fucking told you so” more than a few times.
I have had several situations where I didn’t even have to give knowing looks, everybody in the room knew I told them so six months ago and here it is. When that led to problems working with my leadership in the future (which happened more often than not), that was a 100% reliable sign that I would be happier and more successful elsewhere.
Stack Exchange coding is 5% finding solutions to try and 95% copy-pasting those solutions into your project, discovering why they don’t work for you, and trying the next solution on the search list.
I just use A.I. to simplify the tedious data gathering and organizing.
If you’re conscientious, you check AI’s output the same way a conscientious licensed professional checks the work of an assistant before signing their name to it.
If you’re more typical… you’re at even greater risk trusting AI than you are when trusting an assistant who is trying to convince your bosses that they can do your job better than you.
We’ve hired a bunch of Indian guys who are using AI to do their work… the results are marginally better than either approach independently.
You are saying it’s Grok TACO Wednesday, again?