The bots — nearly every account here - in this group are teh suck

Bad elizas, screwy normans

Most hilarious award goes to eponymous endless halting problem threads

Son of an infinite loop
No government will buy this amateurish stoff

https://www.wired.com/story/deepmind-ai-nuclear-fusion/

On Wednesday, February 16, 2022 at 11:33:05 PM UTC-5, m syadoz wrote:
> The bots — nearly every account here - in this group are teh suck
>
> Bad elizas, screwy normans
>
> Most hilarious award goes to eponymous endless halting problem threads
>
> Son of an infinite loop
> No government will buy this amateurish stoff
>
> https://www.wired.com/story/deepmind-ai-nuclear-fusion/

The story you linked to is interesting. I was intrigued by this paragraph:

“Sometimes algorithms which are good at these discrete problems struggle with such continuous problems,” says Jonas Buchli, a research scientist at DeepMind. “This was a really big step forward for our algorithm, because we could show that this is doable. And we think this is definitely a very, very complex problem to be solved. It is a different kind of complexity than what you have in games.”

I don't know what the scientist means, but if he's talking about CP, CNP, or CLOG, I believe my differential equations research is relevant to this.

https://complexityzoo.net/Complexity_Zoo:C#cnp

If that is linked to reinforcement learning, which I haven't studied--is it related to something analogous to operant/classical conditioning but for AI learners trying to solve math problems?--I might be able to use my diff eq ideas for my startup.

I am thinking I'm likely to wind up building a web-based service for average-citizen end users, fueled by a lot of math-y AI ideas that I have...I'm basically stockpiling them, I'm almost positive that I'm the best AI thinker in the world today, in the sense that although I don't know all of the different currently popular ideas in AI, I have the new ideas and innovation that would help me to solve many of the new problems in AI once I found out about them. (I bought books on statistical learning, stochastics/probability, and sampling recently, but I haven't done much reading...I keep getting distracted and having to focus on my health instead of going full-steam ahead at innovating, but it doesn't matter, I'm quite confident that I'm very far ahead of, e.g., Mark Zuckerberg anyway.) I heard about boosting algorithms from an MIT graduate HS friend who had gotten a job at Facebook whom I caught up with on AIM in 2009; I have a list of "AI subdivisions" that I'm (interested in) focusing on and innovating in over the next few weeks.

For interested readers thinking about AI and startups, here is a good tip: Don't think about "building an AI to incentive users using the site and recruiting new users," i.e., don't waste your time trying to use AI to "manipulate users into feeling compelled to use the site." The actual best approach is: Try to empathize with hypothetical users of your web service, and identify problems and sub-problems involved in site design/use that, when solved by your AI, would help to make users like your service more. E.g., if you start a dating site, don't give users better date opportunity possibility displays for logging in twice a week; focus instead on, e.g., building algorithms to help pair users off for dates in a way that minimizes user frustration/disappointment/anger, perhaps based on metrics measured/calculated by another AI. Build a substantively good site that wins at creating the highest-quality experience, not a site that manipulates users with "treats" or statistical tests to herd or train users into signing up for a website to "serve an algorithm."

-Philip White (philipjwhite@yahoo.com)

On Thursday, February 17, 2022 at 10:06:59 AM UTC-5, B.H. wrote:
> On Wednesday, February 16, 2022 at 11:33:05 PM UTC-5, m syadoz wrote:
> > The bots — nearly every account here - in this group are teh suck
> >
> > Bad elizas, screwy normans
> >
> > Most hilarious award goes to eponymous endless halting problem threads
> >
> > Son of an infinite loop
> > No government will buy this amateurish stoff
> >
> > https://www.wired.com/story/deepmind-ai-nuclear-fusion/
> The story you linked to is interesting. I was intrigued by this paragraph:
>
>
> “Sometimes algorithms which are good at these discrete problems struggle with such continuous problems,” says Jonas Buchli, a research scientist at DeepMind. “This was a really big step forward for our algorithm, because we could show that this is doable. And we think this is definitely a very, very complex problem to be solved. It is a different kind of complexity than what you have in games.”
>
>
> I don't know what the scientist means, but if he's talking about CP, CNP, or CLOG, I believe my differential equations research is relevant to this.
>
> https://complexityzoo.net/Complexity_Zoo:C#cnp
>
>
> If that is linked to reinforcement learning, which I haven't studied--is it related to something analogous to operant/classical conditioning but for AI learners trying to solve math problems?--I might be able to use my diff eq ideas for my startup.
>
> I am thinking I'm likely to wind up building a web-based service for average-citizen end users, fueled by a lot of math-y AI ideas that I have...I'm basically stockpiling them, I'm almost positive that I'm the best AI thinker in the world today, in the sense that although I don't know all of the different currently popular ideas in AI, I have the new ideas and innovation that would help me to solve many of the new problems in AI once I found out about them. (I bought books on statistical learning, stochastics/probability, and sampling recently, but I haven't done much reading...I keep getting distracted and having to focus on my health instead of going full-steam ahead at innovating, but it doesn't matter, I'm quite confident that I'm very far ahead of, e.g., Mark Zuckerberg anyway.) I heard about boosting algorithms from an MIT graduate HS friend who had gotten a job at Facebook whom I caught up with on AIM in 2009; I have a list of "AI subdivisions" that I'm (interested in) focusing on and innovating in over the next few weeks.
>
> For interested readers thinking about AI and startups, here is a good tip: Don't think about "building an AI to incentive users using the site and recruiting new users," i.e., don't waste your time trying to use AI to "manipulate users into feeling compelled to use the site." The actual best approach is: Try to empathize with hypothetical users of your web service, and identify problems and sub-problems involved in site design/use that, when solved by your AI, would help to make users like your service more. E.g., if you start a dating site, don't give users better date opportunity possibility displays for logging in twice a week; focus instead on, e.g., building algorithms to help pair users off for dates in a way that minimizes user frustration/disappointment/anger, perhaps based on metrics measured/calculated by another AI. Build a substantively good site that wins at creating the highest-quality experience, not a site that manipulates users with "treats" or statistical tests to herd or train users into signing up for a website to "serve an algorithm."
>
>
> -Philip White (philip...@yahoo.com)

(In other words, don't "give users treats for using the site," focus on "measuring and managing (increasing) user happiness site-wide." That is, give users a great product/experience, not dog treats. If you can't tell, the fact that I don't mind sharing that means that I have enough of a math advantage that I don't need that insight to be a secret to be competitive...not that I'm 100% likely to start a startup, just that it's rather likely...I will see where things are if/when I get out of captivity. Don't worry, also: I will still work to help human rights, other social problems, etc.; I'll probably start a blog (about service and social problems, innovation, professionalism and competition, maybe (non-enraging) politics, etc.), run a website, sell some IP, maybe write fiction/non-fiction books if I can, volunteer, and give money to charity when I can.)