TechCrunch has a great article today about Rodney Brooks thinking that "people are vastly overestimating generative AI." Key points: • Brooks says that people see things that AI systems can do that are similar to what humans can do, and then they optimistically over-generalize. • What generative AI systems are doing is not the way that humans do similar tasks. • Language (LLMs) is not all that useful for tasks that we can apply to AI today. It just slows things down. • Robots that look like shopping cars are going to be more useful than robots that look like humans. Overall, a fun article! https://lnkd.in/ecbckzKQ
Reading closely machine-generated output takes time and energy and is a cognitive load.
Not overestimated when it comes to raising venture capital. The hottest thing since combining diagnosis and therapy.
People become fixated on a particular current aspect of (what we call) AI, then those ideas are inculcated. Not everyone is playing that game.
Team Builder, Startup Cofounder and App Store Inventor
1mo"Robots that look like shopping cars are going to be more useful than robots that look like humans" -- that's for sure, but AI is highly useful in its current form and may take a while to even apply what we have -- but I'd agree, it's not smart or thinking or human at all, I may never be quite the same as us, in fact it certainly won't be. It's FASCINATING to see what will happen and how AI defines our own selves even.