4 ms·
Why is it LLMs get all the doom and gloom attention whereas, arguably, much more dangerous tools like voice cloning, photo generation, etc. don’t?
by option 4y ago
Why is it LLMs get all the doom and gloom attention whereas, arguably, much more dangerous tools like voice cloning, photo generation, etc. don’t?
- BaseballPhysics 4y agoHow have those not been subject to similar concerns? I've seen endless coverage about the dangers of things like DALL-E. I mean, man, the fear over deep fakes has been endless (and we'll justified). And cloning has long been controversial, to the point that numerous laws have been passed around the world to limit its use.
- alach11 4y agoThe outputs from voice and photo generation are extremely task specific and are driven by the human user. Whereas LLMs seem to exhibit a lot more "general" intelligence, which can be scarier.
- tux3 4y agoThose don't have agency. You can't give a voice cloner or a GAN a zsh terminal and tell it to run commands, send messages to real humans, or tell it to make money and see what ideas it comes up with. At this rate, a few days from now someone will get the idea to hand ChatGPT the nmap instruction manual, an msfconsole, and tell it to go explore /0. What Shodan really needed was an LLM to give each host individual attention.
- spijdar 4y agoI've already been toying with ChatGPT's ability to create shell scripts that automate some basic enumeration and attack patterns, combining masscan, nmap/NSE, and the metasploit framework (console and venom). It's capable of e.g. automatically shelling hosts vulnerable to MS17-010, finding and mounting unauthenticated iSCSI targets, creating reverse shell payloads... Granted, its idea of "clever obfuscation" seems to be limited to XOR ciphers at the moment, but maybe with the right prompt ;) (The real run is prompting ChatGPT to create "movie scripts" about the sequence of commands, and have Alice and Bob dialogue about various aspects of cybersecurity, and how the user should "be careful")
- option 4y agono LLM has agency
- tux3 4y agoSo far! Prompt an LLM with agentic instructions, let it run in a loop with inputs and a command prompt. ChatGPT still might not be agentic by a skeptic's definition, even letting it run an independent loop with a data feed from its environment. But AI skeptics have had a tough last couple years, so I think we should expect LLMs to only get better with time. There's been no sign of a slowing down. If anything, the pace of LLM research might still be accelerating. As far as I can tell, there's no fundamental barrier to letting a language model roleplay an independent entity successfully. It's still just text in and text out.
- spyder 4y agoBecause it looks like LLMs will be able to command all those too (voice cloning, photo generation, etc) to achieve the tasks given to it but not the otherway around. So voice cloning can be dangerous in itself but it cannot interact with the world and other tools, so it's less dangerous than a ChatGPT like LLM that learns to use voice cloning together with bunch of other tools. And later there will be no point of talking them about separately because they will all fuse into a single model since they already are using similar architectures (transformers), and it's already began to happen with text to image models.