Browse Comments — Raw (as collected)

The Stack Overflow answers stick with you.

You are right that confidence without correction is dangerous. The comic nails that danger, but it's not how today's models actually work.

Spending billions to create infrastructure that will ultimately reduce a work force which is expected to foot the power bill? Huh?

Hilarious, but so true

Mine confidently diagnosed a risk issue… and recommended mindfulness instead of controls.Still waiting for the audit trail.:-)

Hank Green - Does this change your take on AI and resource use? I'm curious.

The fundamental question here.. Does the man need to ask AI whether he should run? Will human intelligence get so dependent on AI that we cant make even instinctive survival decisions?

Why is Gemini3.5 12X faster in Antigravity IDE?

Reginald Mba AI only amplifies what we can do. It will probably lead us astray if we use it as an expert to get answers to questions in matters where we are still novices😅

That's a really clever approach to engineering, but framing it as 15–25x cheaper than DeepSeek is a bit misleading. Leveraging a cascade of tiny, hyper-focused models makes total sense for optimizing a specific vertical like voice moderation, but I don't think you can compare that efficiency against a generalized foundation model built for massive reasoning tasks. Still, it's a super cool way to solve the compute bottleneck for your specific use case!

AI: I can speak true or false, but what i say, i say with confidence 😁

Well, AI costs are soaring, so I guess you have a choice: solve your issues with the community or max out your credit card.

The confidence point is very recognisable, and it raises something I keep coming back to in my own use of AI. What looks like confidence is not really confidence in the human sense. The model is producing an answer through a mathematical/probabilistic process, not weighing doubt or certainty as a person might. That is why fluency can be so misleading: we can mistake it for judgement, and certainty for truth. Verification matters; so does resisting the temptation to anthropomorphise what the system is doing.

I think this was an episode of black mirror

Love it!... hahaha. Always remember: Human in The Loop, but not in that way...

My take on this is, prompts are not detailed and specific enough. Even if you ask your own father or relative not a detailed question, you get a different answer. There is no point making fun of AI Chatbots Pascal BORNET

"My life experiences told me all the advice AI was giving me on a new career path was too good to be true. So I kept doing my research rather than blindly going along. Sadly, I wonder how many out there without that experience will just go along with what it tells them.

Gordon Anderson CEng MIET The subsidy through infrastructure.

Sonia Felix perhaps the AI answer will be something you already know and will not like.

Sakib Ziad You're comparing apples to oranges. Books, calculators and search engines do not give ready made answers to asked questions. Users were still required to put in the work, go over the articles, analyze and record knowledge and make informed decisions all the while using their brain. But with LLM, there is no effort, hence less learning and less critical thinking. So the orginal argument still stands valid. LLM is making humans dumber.