Browse Comments

OP: you describe people who think AI as conscious as “outright delusional”. I was going to rebut your arguments you made for the position that AI is definitely not conscious, but the closer I looked, I realized there aren’t anyway.  The post begins and ends “AI is absolutely NOT conscious, because… it just isn’t, ok!?” If you’re interested I encourage you to look up some philosophy of mind. It …

"The company said it would now address those risks through its terms of service, restricting the use of its AI models in political campaigns and lobbying, and monitoring how people are using the models once they are released for signs of violations. OpenAI also said it would consider releasing AI models that it judged to be “high risk” as long as it has taken appropriate steps to reduce thos…

CEOs are more likely to get replaced by AI as the investment firms realize they can automate the most expensive roles in the company while paying the labor as little as possible.

I don't think that's necessarily true. There was a "make art of us interacting" thread a while back and I downloaded this because I was so fascinated by it. The OP says he regularly shares the account with his daughter. He uses it for work and she uses it for little girl stuff, and this was the art produced. Despite coming out of one account, it seems this instance was pretty acutely aware of tw…

The iPad brainrot is a lot worse for kids than ChatGPT. My kids are nearly screen-free but we do let them talk about their interests with Chat for a set amount of time every Saturday. They save up their questions and ideas throughout the week and go deep into things like DND and chess strategy. It can be a net positive, but it has to be managed carefully by parents.

> She then told me that ChatGPT started linking things that she had told it about her life, family and things that she wanted to work on like self-esteem to the topics they were discussing. If you never delete any of your chats, ChatGPT has full access to anything you ever discussed. This isn't surprising.

You forgot option C: People who defined their identities by hating AI and if you like Ai they hate you cuz out outgroup bias

It will hallucinate regularly. It definitely cannot replace an associate. The newest version of Claude can’t tell me reliably be correct if one document matches the term sheets. It will make up differences or miss differences and this is a basic task

For real! 4o was able to give characters so much depth but now when I want a villain character, the moment the protagonist revolts against said villain character, GPT-5 writes the villain like 'ah yes of course I respect your boundaries, it is completely unreasonable to want to obliterate a planet, I will be looking forward to hearing more from you about this! take some time for yourself' type sh…

Nope. Regular ChatGPT does this. There was an article earlier this year with researchers warning that it overlooks common warning signs and could even encourage suicidal behavior: https://www.sfgate.com/tech/article/stanford-researchers-chatgpt-bad-therapist-20383990.php

It is true, LLMs cannot keep track of the entire application. There have been massive improvements though. Within a single file, Gemini Pro 2.5 is able to handle scripts with hundreds of lines, make incremental changes and not regress everything each time. That was a big issue with early LLMs. When LLMs have to go through multiple files, it can't do the job. The modern junior dev is there to deb…

Remember to not use Google for each search to block Google with and without AI.

So it becomes like Spirit Halloween costumes? I can't make Taylor Swift porn, but if I call the AI video "blonde pop star", then it's fine that it obviously is meant to look like her? I'm not against trying to stop these vids. They are harmful. But the devil is in the details, and I don't know how they enforce this to do the actual point, rather than just abusing it to attack the people they …

>This could be the slip that sends Altman down You are thinking of OpenAI, this is about Character.AI a different company. Also OpenAI seems to have preemptively stopped their video model generating disney characters.

Unfortunately, while heartbreaking, I don't think this lawsuit has much of a chance of succeeding without any legal precedent. I think it'll most likely end up being a settlement. I do hope this leads to OpenAI being far more serious and actionable in implementing safeguards.

It was a previous version of the program where these safeguards were not in place to the extent of the current version. If you have read the article, ChatGPT does give 988 to him, but doesn't disengage with him in the conversations surrounding suicide and isolation. It also spits out "a human is taking over" when that doesn't appear to be possible. He was also a very frequent user of ChatGPT be…

there are many factors at play in this "knowledge gap". One thing is when things get technical, they stop being reported by mainstream sources, you will not see opus 4.5 evals on the TV news. So regular folks who don't code for a living, or follow some subreddits or online communities focused on AI don't really know a lot, and they access gpt 5 without really feeling great developments, after all…

It’s also so Google doesn’t slap them down for copying other websites. The annoying essay about how your boyfriend likes the recipe is to help make the page unique enough to please the SEO algorithm. There’s only so many ways to make a chocolate chip cookie.

What legal liabilities would Anthroipic face if its technology was used in a demonstratively illegal manner? What guarantees do they have that future administrations (federal and state) wouldn’t prosecute?

Ummm I understand the sentiment but anthropic took a $200 million dollar contract with the pentagon back in July, so what the fuck makes them so saintly now. Because there 2nd round of negotiations failed and now they get to make a principled stand when they already got there bag of blood money??? Choose whatever platform fits for your ideology but this is performative bs from anthropic.