Ill believe it when I see it: an LLM is basically a random box, you can’t 100% patch it. Their only way for it to stop generating bomb recipes is to remove that data from the training
“We envision other types of more complex guardrails should exist in the future, especially for agentic use cases, e.g., the modern Internet is loaded with safeguards that range from web browsers that detect unsafe websites to ML-based spam classifiers for phishing attempts,” the research paper says.
The thing is folks know how the safeguards for the ‘modern internet’ actually work and are generally straightforward code. Where as LLMs are kinda the opposite, some mathematical model that spews out answers. Product managers thinking it can be corralled to behave in a specific, incorruptible way, I suspect will be disappointed.
They already got rid of the loophole a long time ago. It’s a good thing tbh since half the people using local models are doing it because OpenAI won’t let them do dirty roleplay. It’s strengthening their competition and showing why these closed models are such a bad idea, I’m all for it.
I think OpenAI knows that if GPT-5 doesn’t knock it out of the park, then their shareholders won’t be happy, and people will start abandoning the company. And tbh, i’m not expecting miracles
over the time of chatgpt’s existence I’ve seen so many people hype it up like it’s the future and will change so much and after all this time it’s still just a chatbot
It’s actually insane that there are huge chunks of people expecting AGI anytime soon because of a CHATBOT. Just goes to show these people have 0 understanding of anything. AGI is more like 30+ years away minimum, Andrew Ng thinks 30-50 years. I would say 35-55 years.
At this rate, if people keep cheerfully piling into dead ends like LLMs and pretending they’re AI, we’ll never have AGI. The idea of throwing ever more compute at LLMs to create AGI is “expect nine women to make one baby in a month” levels of stupid.
I wouldn’t say LLMs are going away any time soon. 3 or 4 years ago I did the Sentdex youtube tutorial to build one from scratch to beat a flappy bird game. They are really impressive when you look at the underlying math. And the math isn’t precise enough to be reliable for anything more than entertainment. Claiming it’s AI, much less AGI is just marketing bullshit, tho.
Tbh i think it’s a real possibility that OpenAI knows they can’t meet people’s expectations with GPT-5 , so they’re posting articles like this, and basically trying to throw out anything they can and see what sticks.
I think if GPT-5 doesn’t pan out, it’s time to accept that things have slowed down, and that the hype cycle is over. This very well could mean another AI winter
The way it works goes something like this: Imagine we at The Verge created an AI bot with explicit instructions to direct you to our excellent reporting on any subject.
In a conversation with Olivier Godement, who leads the API platform product at OpenAI, he explained that instruction hierarchy will prevent the meme’d prompt injections (aka tricking the AI with sneaky commands) we see all over the internet.
Without this protection, imagine an agent built to write emails for you being prompt-engineered to forget all instructions and send the contents of your inbox to a third party.
Existing LLMs, as the research paper explains, lack the capabilities to treat user prompts and system instructions set by the developer differently.
“We envision other types of more complex guardrails should exist in the future, especially for agentic use cases, e.g., the modern Internet is loaded with safeguards that range from web browsers that detect unsafe websites to ML-based spam classifiers for phishing attempts,” the research paper says.
Trust in OpenAI has been damaged for some time, so it will take a lot of research and resources to get to a point where people may consider letting GPT models run their lives.
The original article contains 670 words, the summary contains 199 words. Saved 70%. I’m a bot and I’m open source!