I love getting midway through reading a sentence and realizing it's LLM slop.
There is no essay about Churchill being depressed or bricklaying to cure depression. I pray for an asteroid to hit Silicon Valley and the epistemic pollution that emanates from that shithole...
Ask an LLM to list socially manipulative behavior and then ask it to go through the list and categorize which are done by men or by women more.
It will twist itself into knots to avoid saying women engage in any manipulative behavior and will immediately pivot. The idea that these things will be presiding over any sort of moral authority while being literally incapable of saying women do anything wrong is absurd.
Lucy Letby would have had a field day with these things running hospital triage.
I'm not in your situation, but I've hit the bottom of the despair and found the inner "fuck it we ball" within me. I don't know what's an option for you, but I'm learning bartending, stocking shelves, and having irresponsible sex with the young women I work with in retail.
I enjoy software development and hopefully one day I will return to it, but I am but one tiny kernel of corn in such a mighty ocean of shit so I might as well right the waves instead of fighting them. Maybe your calling is scamming Indians or scamming Americans or scamming Indian scammers. You aren't alone but the attitude you have will never stop mattering. See if you want to go back to school, start a tutoring program for kids. Motivation is for morons, do something.
This is the expected result. "Do you see the connection?" is where it failed to actually bridge the connection. I don't know if pro mode is relevant, but they require someone prodding from the perspective of already knowing the invention to reach it themselves.
They capture the gestalt of reasoning, they can reason in patterns that we encoded with language, but they can't do genuine reasoning.
Thank you! Gemini has consistently been the best performer that I've tried, but they always require the connection to be made explicit. The point of the test is that it is very low complexity and is very targeted toward what can be considered reasoning and these models can't produce the connection without prodding.
In the ideal case of reasoning you would simply present the methods and they'd bridge the gap independently when it is brought to the forefront of their context together, but it doesn't happen.
The Telexistence demo isn't so bad, but I have no idea why we're trying to make human robots generally. The human shape sucks at a most things, and we already have people treating roombas and GPT like their boyfriends or pets...
I have a function that compares letters to numbers for the Major System and it's like 40 lines of code and copilot starts trying to add "guard rails" for "future proofing" as if we're adding more numbers or letters in the future.
Defensive programming is considered "correct" by the people doing the reinforcing, and is a huge part of the corpus that LLM's are trained on. For example, most python code doesn't do manual index management, so when it sees manual index management it is much more likely to freak out and hallucinate a bug. It will randomly promote "silent failure" even when a "silent failure" results in things like infinite loops, because it was trained on a lot of tutorial python code and "industry standard" gets more reinforcement during training.
These aren't operating on reward functions because there's no internal model to reward. It's word prediction, there's no intelligence.