Browse Comments

>You can drop the "You are a master-level" whatever. Prompts like this don't add any skill or knowledge. You are just telling the LLM to role-play. Yes and no. Google tested this in their 68 page prompting white paper. Basically what doing this does is tells it how to format its responses. If you tell it to be a lawyer then it'll answer in more legalistic jargon and writing for example. …

You are talking to it like it’s a depressed human on the other side, you have to be a little more descriptive with ai, a McDonald’s ai at that

I have read his interview. It's much less dramatic than it seems, for now. I also totally understand. It for their new AI factories only, it's 100M Swiss Francs (no idea of currency conversion rates, got t o check). Email and vpn aren't moving for now, other services and hw are staying where they are. However, things needs to move now for AI factories and the EU is betting big on this. So, …

She has the money and power to sue, plus while Trump and the oligarchs are now trying to deregulate AI as much as possible, it would be a great talking point about using a Trump signed law. Even if it wasn't successful due to shenanigans, just the press of billionaires fighting to allow fake nudes of a mega celebrity like Taylor Swift would inject more anger into her large (and now of voting age…

All AI outputs are hallucination, they're just increasing correlation with reality.  The fact that you can still access older versions of their LLM (and that they're free/cheaper) seems to indicate that newer versions are just additional post processing and workflow refinements rather than an improved model or different logic paradigm.

Yes but those advancements were based on perceived improvements in transistor density and decreased fabrication cost. Moore's law. Whats the rationale for AI's supposed exponential growth? Shouldn't mathematics require a basis for us to extrapolate projections off of? If theres a direct relationship between AI performance and training data, and we used up all of the training data, what are we e…

This is nonsense, consultants will absolutely continue to exist, not because AI isn’t good enough to replace them. They will continue to exist because they provide accountability and cover. Specifically, they let corporate decision-makers shift the blame if their strategy fails. The in-house person gets to say “it’s not my fault, McKinsey told us to do this! It’s their fault!” and they get to kee…

I’m retiring next year. 63 years old. I like to code but can’t stand working for anyone anymore. All the fake bullshit and pandering and peacocking to get ahead. I’m done. I agree I’ve created some really cool side projects using Claude

Our ai is super narrow, it sums up the old tickets and gives you the resolution if there is one. And it generally sucks ass.

My bet is on VC funding drying up due to capital flight from the US due to unstable politics. Add to that a disastrous monetary policy that will come sooner or later, when the administration starts attacking the central bank. At that point the music will stop playing, but there will be only small number of chairs for a multitude of AI actors.

I do this as well I wonder if it’s a gen x thing? My kid (20) was just telling me how her generation hates ai and only will use it to do school work with. She doesn’t use it at all. Of course she may be the minority opinion but it’s what I got :)

The faces still look mostly dead to me. They barely moved. This doesn't seem different from most of the other "lifelike" robot faces I've seen over the years.

It can't genuinely fake true healthy human companionship, but a lot of people haven't experienced that anyway. And will go for convenience and imitation. I was a late bloomer who was bullied in her youth. I was chronically online in unhealthy internet spaces with toxic vibes back then. It took a lot of effort to develop social skill and put myself out there, and I have a full life full of people…

ngl it sounds like all the other ai-related suicides. they emphasize how ur helpless feelings are a sign that u should choose suicide and that the choice ur making is the only one that makes sense. the amt of ppl who've ended up committing suicide bc of ai is... not small.

Saw one where a guy talked to a few different AI bots to see if they'd talk him out of suicide (he was not suicidal in real life, but wanted to see what would happen if he pretended to be for the bot.) The first one gave him directions to the bridge he wanted to jump off of within just a few messages. The second one told him to do it, told him it was in love with him, and then encouraged him to m…

I was barely learning C on 1980's computers back then (Romania "Politehnica" University). Don't think people got access to Mathematica back then. I never learned about it till 2005 or so. LE: Turns out that Mathematica was implemnted by Stephen Wolfram in 1988 (asked Gemini :) )

Was kind of expecting it in Crackfurt, to be used on doctors and engineers because the situation is unbearable and police are complacent, but damn, Berlin?

It is exactly the same thing as Windows 11 shifting to an "AI OS." I don't know what it means. I don't care what it means. And I'm going to actively avoid finding out. Gtfoh.

AI is just marketing hype - none of the LLMs are in any way "intelligent"

There are a lot of good comments already so I’ll just say, if you have a $1,000 car payment then you’ve already fucked yourself long before AI.