I liked this footnote. A simple, intuitive explanation.
> If you pay $1.00 to press a button, and pressing the button pays you $0.99, you > will press the button zero times. If you get $1.01 instead, you will press the > button all the time.
One thought: If it only gives you $0.01 profit you'll (have to) push it more than if it gave you $1000 profit. There's a saturation point.
The ultrasound technique there is more like MRI, static imaging, not measuring dynamic electrical signals. Also, regardless of static vs moving, all of this hinges on massive relatively expensive devices (ultrasound is never going to be dirt cheap or miniaturized compared to a smartwatch or even a VR headset) and for the subject to remain perfectly still and probably go through regular calibration sequences. All for <75% accuracy on simple classification tasks. The information mixes in the propagation out of the skull, erasing information. It's analogous to trying to do a row hammer attack on a CPU from outside the computer case.
Also, in the Meta result here, "while actively typing" is actually quite different than passive mind reading because the motor cortex sits nicely near the top surface of the skull, and the muscle memory from past typing makes for a nice well formed signal to measure and classify. It's the same trick over and over since the Brown BrainGate days where you can have people perform or imagine movements and get a decent but not good classification result, and it never gets much better after that trick is exploited. Project dies, VCs and grant writers forget or never appreciated the effect, time goes by, a grad student or corporate research lackey rediscovers it, media puts out an article claiming mind reading is here, and the cycle repeats...
It’s more of a testament to a potential beneficial side effect of an already widely used drug for something entirely different.
I (in America) actually exercise 5-6 days a week with regular sauna use afterwards averaging about 2-2.5 hours per total session (20 minutes from sauna). I coincidentally even happen to take long hot baths frequently (3-4 times a week for an hour) as a form of stress relief, relaxation, and it further helps with muscle recovery.
Absolutely none of that helps me to a notable degree with what the specific pill in question does which I also use regularly. In addition, viagra adds an additional say 4 hours per day per pill to my normal weekly routine of increased blood flow that may help reduce risk of dementia when I’m older. That’s a big win in my book if it’s true as it’s something I already do and is additive.
Have done regular exercise to improve my Long COVID state for 6 months now. It has helped, but it has its limits too. If I overdo it, I get worse, but I can't always tell until later. And other health issues are limiting what forms of exercise I can do.
Hard problems rarely have one simple solution, or they wouldn't be hard, would they?
What’s wrong with being excited and working hard towards something? Hoping for nearly anyone’s demoralization is not good. I wish you a spirit of adventure that takes you many exciting places!
Also, what specific safety problems do you think need to be solved?
There’s a sudden market for a new flavor of bullshit, and brand-new bullshit providers have sprung up to fill the vacuum. But all the exec-webinar/thought-leader/hypetrain/pr stuff is a sideshow from people building actual value.
> What’s wrong with being excited and working hard towards something?
Ahh, the old "how can cryptocurrency be harmful if people like buying it?" argument.
There are many realistic and amenable goals in life, like mapping the human genome or writing an Open Source microkernel. Declaring that you intend to surpass human intelligence via a statistical text generator is not one of those things. It does not have precedent, it does not have feasibility studies, it does not seemingly indicate in any way shape or form that it is possible. Nobody can even spell-out the intermediate steps to get us to AGI; every single "novel" solution involves scaling up our current, broken, concepts. It's ELIZA versus the Lisp pundits all over again.
Excitement and hard work go a long ways, but you won't know which way until you apply a little logic. The current "AGI" trend is practically non-existent outside the venture-capital sphere and OpenAI employees, both of whom would be bullish on AGI anyways since it's good for business. Once you discard the biased opinions, you're left with legitimately confused investors and nonsense opinions propagated by conspiracy theorists. Much like crypto, the concept of "AGI" is being used to confuse and exploit people who misunderstand technology and finance.
> Also, what specific safety problems do you think need to be solved?
I think you misread their comment. They said "safely incorporated", which is not a specific safety problem for AI but a holistic consideration that stops your product from sucking. Your computer vision model could be statistically perfect, but absolutely useless for self-driving tasks and multimodal robotic agents. It's not taste that separates these good implementations from the bad ones, it's logic. You have to be considerate when implementing AI in traditional systems, because inherently AI can be wrong and you must have a failure-mode for those situations.
Many people reject this idea, because it precludes the idea that someone could sell a cure-all to today's AI ills. But real AI safety cannot be baked-in to a model. It only exists when genuinely thoughtful humans anticipate every single fail-state; if that sounds like hard work, it's because it is. And nobody, nobody, sells it as "AGI".
> If you pay $1.00 to press a button, and pressing the button pays you $0.99, you > will press the button zero times. If you get $1.01 instead, you will press the > button all the time.
One thought: If it only gives you $0.01 profit you'll (have to) push it more than if it gave you $1000 profit. There's a saturation point.