I don’t think the smart people saying it’s too smart are using the right adjective, but they’re also not wrong.
Let’s say AI makes a hallucination, which it’s prone to do, and decides that it’s absolutely necessary to gain access to the power grid, for whatever reason which isn’t important. It then uses advanced, sophisticated techniques to gain access to their network, and in doing so accidentally takes out power to a large area. The agen was absolutely being intelligent, it managed to get into a protected network, but it wasn’t exactly being smart, as it was clumsy and unrelated to the task. Now extend that to an agent who’s been instructed to find a solution at all costs, and the hallucinations and pervasiveness get far more serious.
I too initially thought that yeah AI is too stupid to cause actual disasters, but thinking about it properly, i don’t think it is.
This is some American tech oligarch propaganda. They use this lie to get all FOSS models banned, because IT’S SO DANGEROUS. Then, they are the monopoly on text generators and can charge what ever they want to keep the round-tripping scam going. Never believe a doomsday cult.
This is exactly the lie that scam peddlers are telling. There is no “agent” that is so very intelligent it can do something that 10 script kiddies with access to internet and too much free time couldn’t do already. An “agent” is a python script, it doesn’t have intent, agency, intellect, or any other qualities we like to ascribe to inanimate objects, and just because llm can imitate speech it doesn’t make it more so.
Hmmmm script kiddies can do a lot of damage tho. And if the model is on that level, they can as well. Plus, unlike script kiddies, the robot never tires, gets distracted, or bored, or hungry, making it capable of utilising that destructive power at way higher speeds.
Script kiddies never did their shenanigans manually. And they were always perfectly capable of running their scripts in a while loop. A “robot” that you’re describing is the same script in a loop, running on the same machine, but instead of running their scripts from a static file, it now runs it from the output of an LLM, and even though sometimes it yields some unexpected results, on average it’s nothing new.
The difference is, a script that tries known exploits against a list of IP addresses and ports costs nothing. Same, but put through an LLM costs money, for worse “results”.
I feel like you underestimate both script kiddies and the bots. I could do a lot worse when i was 13 than what youre describing, and ive seen bots do similar harm.
I don’t think the smart people saying it’s too smart are using the right adjective, but they’re also not wrong.
Let’s say AI makes a hallucination, which it’s prone to do, and decides that it’s absolutely necessary to gain access to the power grid, for whatever reason which isn’t important. It then uses advanced, sophisticated techniques to gain access to their network, and in doing so accidentally takes out power to a large area. The agen was absolutely being intelligent, it managed to get into a protected network, but it wasn’t exactly being smart, as it was clumsy and unrelated to the task. Now extend that to an agent who’s been instructed to find a solution at all costs, and the hallucinations and pervasiveness get far more serious.
I too initially thought that yeah AI is too stupid to cause actual disasters, but thinking about it properly, i don’t think it is.
This is some American tech oligarch propaganda. They use this lie to get all FOSS models banned, because IT’S SO DANGEROUS. Then, they are the monopoly on text generators and can charge what ever they want to keep the round-tripping scam going. Never believe a doomsday cult.
This is exactly the lie that scam peddlers are telling. There is no “agent” that is so very intelligent it can do something that 10 script kiddies with access to internet and too much free time couldn’t do already. An “agent” is a python script, it doesn’t have intent, agency, intellect, or any other qualities we like to ascribe to inanimate objects, and just because llm can imitate speech it doesn’t make it more so.
Hmmmm script kiddies can do a lot of damage tho. And if the model is on that level, they can as well. Plus, unlike script kiddies, the robot never tires, gets distracted, or bored, or hungry, making it capable of utilising that destructive power at way higher speeds.
Script kiddies never did their shenanigans manually. And they were always perfectly capable of running their scripts in a
whileloop. A “robot” that you’re describing is the same script in a loop, running on the same machine, but instead of running their scripts from a static file, it now runs it from the output of an LLM, and even though sometimes it yields some unexpected results, on average it’s nothing new.The difference is, a script that tries known exploits against a list of IP addresses and ports costs nothing. Same, but put through an LLM costs money, for worse “results”.
I feel like you underestimate both script kiddies and the bots. I could do a lot worse when i was 13 than what youre describing, and ive seen bots do similar harm.
Not really, I agree that they both bring harm, my point is that LLM bots aren’t that much more dangerous, if any.
i think you’re giving AI’s capabilities a lot of credit