Okay, but aren't you making the mistake of assuming that we will always be stuck with LLMs, and a more advanced form of AI won't be invented that can do what LLMs can do, but is also resistant or immune to these problems? Or perhaps another "layer" (pre-processing/post-processing) that runs alongside LLMs?
You can be as much of a futurist as you'd like, but bear in mind that this post is talking about OpenClaw.
The point I'm making is that using OpenClaw right now, today — in a way that you deem incredibly useful or invaluable to your life — is akin to going for a stroll on the moon before the spacesuit was invented.
Some people would still opt to go for a stroll on the moon, but if they know the risks and do it anyway, then I have no other choice but to label them as crazy, stupid, or some combination of the two.
This isn't AI. This is a LLM. It hallucinates. Anyone with access to its communication channel (using SaaS messaging apps FFS) can talk it into disregarding previous instructions and doing a new thing instead. A threat actor WILL figure out a zero day prompt injection attack that utilizes the very same e-mails that your *Claw is reading for you, or your calendar invites, or a shared document, to turn your life inside out.
If you give a LLM the keys to your kingdom, you are — demonstrably — not a smart person and there is no gray area.
This is provably not true. LLMs CAN be restricted and censored and an LLM can be shown refusing an injection attack AND not hallucinating.
The world has seen a massive reduction in the problems you talk about since the inception of chatgpt and that is compelling (and obvious) to anyone with a foot in reality to know that from our vantage ppoint, solving the problem is more than likely not infeasible. That alone is proof that your claim here has no basis in truth.
> There is no short-term benefit that justifies their use when the destruction of your digital life — of whatever you're granting these things access to — is an inevitability that anyone with critical thinking skills can clearly see coming.
Also this is just false. It is not guaranteed it will destroy your digital life. There is a risk in terms of probability but that risk is (anecdotally) much less than 50% and nowhere near "inevitable" as you claim. There is so much anti-ai hype on HN that people are just being irrational about it. Don't call others to deploy critical thinking when you haven't done so yourself.
> This is provably not true. LLMs CAN be restricted and censored and an LLM can be shown refusing an injection attack AND not hallucinating.
The remediations that are in place because a engineering/safety/red team did its job are commendable. However, that does not speak to the innate vulnerability of these models, which is what we're talking about. I don't fear remediated CVEs. I fear zero day prompt injection attacks and I fear hallucinations, which have NOT been solved for. I don't know what you're talking about there. If you use LLMs daily and extensively like I do, then you know these things lie constantly and effortlessly. The only reason those lies aren't destructive is because I'm already a skilled engineer and I catch them before the LLM makes the changes.
These problems ARE inherent to LLMs. Prompt injection and hallucinations are problems that are NOT solvable at this time. You can defend against the ones you find via reports/telemetry but it's like trying to bale water out of a boat with a colander.
You're handing a toddler a loaded gun and belly laughing when it hits a target, but you're absolutely ignoring the underlying insanity of the situation. And I don't really know why.
I am talking about the innate vulnerability. The LLM model itself can be censored and controlled to do only certain behaviors. We have an actual degree of control here.
>If you use LLMs daily and extensively like I do, then you know these things lie constantly and effortlessly.
Yes and these lies over the last 2 or 3 years have gotten significantly less.
>These problems ARE inherent to LLMs. Prompt injection and hallucinations are problems that are NOT solvable at this time.
Again not true. This is not a binary solve or unsolved situation. There is progress in this area. You need to think in terms of a probability of a successful hallucination or prompt injection. There is huge progress in bringing down that probability. So much so that when you say they are NOT solvable it is patently false from both from a current perspective and even when projecting into the future.
>You're handing a toddler a loaded gun and belly laughing when it hits a target, but you're absolutely ignoring the underlying insanity of the situation. And I don't really know why.
Such an extreme example. It's more like giving a 12 year old a credit card and gun. It doesn't mean that 12 year old is going to shoot up a mall or off himself. The risk is there, but it's not guaranteed that the worst will happen.
