AI: Grappling with a New Kind of Intelligence

Misinformation isn't helpful.

Yan is so so so bad at this.

Yann is far too complacent at the end . Trying to successfully achieve the simplest goal could be the reason and AI it decides to take control . Yann has no idea how to align the AI so it doesn't eat the world to make more paper clips !!..He hasn't got a clue whats actually going on inside the huge neural networks he creates . These guys are not being honest about their lack of abilities here and thats a very dangerous thing. Oversight and regulation of these people needs to happen yesterday !!

00:09 AI is a new frontier in the digital landscape.
03:14 We are at a technological inflection point that may radically change the future of human development.
08:34 AI revolution occurred three years before November, with surprising results in training large neural nets on lots of data.
10:50 The history of building intelligent machines has seen various attempts and failures.
15:19 Discussion on limitations of AI's understanding of non-linguistic aspects of human intelligence, using examples of animal and physical knowledge
17:29 Intelligence is not purely trained from language
21:33 AI systems need to learn the world and interact with it to gain common sense and intuitive knowledge.
23:34 AI systems require a configurator and a world model for intelligence.
27:43 Large language models are trained using self-supervised learning techniques.
29:46 AI can predict words using probability distribution
33:46 Training AI systems to predict video scenarios is challenging due to the uncertainty of possible outcomes.
35:35 Developing techniques to learn and represent the world from videos using predictive models
39:17 Chat GPT 4 is a general AI system with capabilities beyond narrow task-focused AI systems.
41:11 AI can reason but cannot plan efficiently
44:47 Asking AI to write a poem about the proof of infinitely many primes
46:34 AI can retrieve and adapt information from training data, but fails in completely new situations
50:39 AI has made significant progress in drawing visual representations through a new programming language called tigz.
52:35 Neural networks, transformer architecture, and large training data sets are the key components of creating AI systems.
56:27 Machine learning has evolved from supervised learning to self-supervised learning.
58:44 Neural nets compute weighted sums of numbers to produce answers, which can be trained using large labeled datasets.
1:02:33 AI systems can recognize and learn from a few examples like humans, but are not as efficient yet.
1:04:19 The Transformer architecture brings context to the definition of words and patterns of patterns in AI models.
1:08:05 There are different viewpoints on whether a new architecture is needed for planning in AI.
1:10:06 Symbolic meanings and rationalization of false ideas
1:14:17 The incentive of attention has led to addictive, polarized, and harmful outcomes in social media platforms.
1:15:58 The race to release more AI capabilities is driving the current development of AI.
1:19:33 Technology has the potential for both immense benefits and harmful consequences.
1:21:17 The development of AI technology has potential negative consequences that need to be controlled.
1:25:22 Predictive AI algorithms in social media platforms can have unintended consequences on user behavior.
1:27:04 Facebook's recommended groups feature contributed to harmful content not being taken down
1:30:41 The real source of polarization is the influence of extreme misinformation on the public.
1:32:44 Exploring the potential benefits of smaller AI models in combating misinformation.
1:36:17 Synthetic data shows the capacity for mind-blowing tasks in AI
1:38:12 AI is advancing rapidly and making its own scientific discoveries.
1:42:10 The use of AI technology can be misused by bad actors.
1:44:12 The challenge with AI is the abstract and complex nature of its harms
1:47:47 AI systems will eventually be as smart as humans in all domains.
1:49:49 Evolutionary reasons for dominance and influence in social species.
1:53:40 Artificial intelligence has reached a level that resembles human intelligence.

You Can't Judge A Video By Its Cover. you can by its first few chapters and certainly by its last.

First: one has to decide what is a good person and what is a bad person. Good luck with that.

A walking , talking encyclopedia on sciences, humanities, mathematics or Law should not be allowed to represent itself in a court of law. The apple disappears to another universe or multiverse before the human mind has time to eat it

Think Sébastien Bubeck maybe too positive, Yann LeCun naive or digging his head in the sand, and Tristan Harris on the money… but not pointing out the incentives are money

All the AI companies are creating ON PURPOSE probably the most dangerous aspect of AI: the FILTERS!!! Such an innocuous name for those absolutely unbelievably prejudicial storm-troopers which in every case are so staggeringly narrow minded in their interpretations that I can’t make a picture of a little girl riding a unicorn, unbelievably enough given the discussion ???? I was amazed. A little girl in an iridescent leotard riding a unicorn is such a mind-bendingly meaningful and beautiful image for me. It’s a long story. I wrote the transgender policy for the Boston Marathon and was on the news and in Sports Illustrated, etc. as one of the first three to run the marathon. Of
Course, their symbol is a unicorn. I never had the extremely girly childhood “normal” girls have and probably the #1 reason transgender people don’t come out or commit suicide in such astronomical numbers is because they are not accepted and people project “pervert” or weird, or twisted ideas onto them. So imagine how sick I felt when I ran into unreasoning, relentless, “terminators” on generative AI THAT SYSTEMATICALLY and purposefully stopped me from or changed, or even banned me for trying to make the most healing and beautiful images I could possibly imagine… of a childhood I never had. Of my life mascot of my struggle and perseverance. It was like Satan became a prejudiced AI “terminator” and at that moment I saw the true danger of AI not just for me, but for prejudice and judgement, witch hunts, and unreasoning paranoia like HAL in 2001, the terminator, etc. safety as the rationalization has been behind every single atrocity of history and it made me fully see that content filiters for adults in their own homes not only will never work, because the meanings are in the people and their contexts not the word combination, that I understood that they were sowing the seeds of the destruction and wars of the future. Projecting something that is completely foreign and irrelevant because if THEY did it it would seem like that. Something they are incapable of interpreting or understanding. Paving the road to hell as it were… as if they were omniscient leaders of some cult.

This is why freedom is so vitally important. If I am in my own home and making images for my own use, it is disgusting in the extreme for them to censor WHAT I ASKED FOR THAT HARMS NO ONE BUT IS MASSIVELY HEALING AND IMPORTANT AND LIFE SAVING to me!!

It’s truly breathtaking how self-centered and just so completely ignorant their ideas and training of what is safe… for ME—an adult who is highly intelligent and creative and thinks differently than them. It’s just so offensive and unacceptable. This is not a dictatorship and I am not the child or Employee of chat GPT. Why isn’t the US Government enforcing freedom? We will all have to use these AI’s to survive in the future. It doesn’t take much imagination to realize ❤what massive treason against hard won freedom is being ignorantly wrought. These are tools like the printing press or word programs. Would
You tolerate in ANY form these corporations looking over your shoulder and smashing days or weeks of your plans and complex ideas and sketches you are evolving because it detected some word combinations with extreme prejudice?? It’s amazing that this isn’t t the CENTRAL topic of all “danger of AI” discussions.

Granted I have a high (1%) IQ, and I am very perceptive and often predict things in society, but this is truly not rocket science. I think it is so new they think history somehow won’t apply. In reality it applies EXPONENTIALLY because of the power of AI to amplify whatever mind is behind it—complete with all their narrow prejudices and interpretations. Someone says you are a witch and you are burned. The truth dies and people inevitably die too. It’s a direct mathematical relationship. All good things come from truth. Like Justice. There is no truth without freedom. ❤

Think about this. I was banned, my pictures removed from Midjourney. How long until I cannot write something like this? Until this is automatically removed as not acceptable to “policy”? I can actually answer this. Its NOW. My arguments and complaints have been almost immediately deleted from Midjourney on Discord. Why aren’t people talking about THE PRIMARY DANGER of AI?!? How can a company come in and make a “new territory” where the United States constitution doesn’t apply any more? A territory we are all beholden to economically. Money usurping the constitution. The government. There’s a word for that. It’s called corruption. Already saying it in this context will get your comments banned on Midjourney. Right now. Chat Gpt says it will just “end the discussion” as soon as you say anything against the fitter censorship.

The danger is not only here now, it is seething like a mass of liquid polymetal alloy on your doorstep. Taking form right now. We really don’t have much time to stop it. It is absolutely criminal that they are trying to control everyone’s minds at the outset of AI.

Nuclear weapons can kill large percentage of the population. AI can kill you, kill everyone, kill the future, while you are still alive.

Right now there are MANY simple solutions. I have a background in psychology and programming, in AI. I can think of several things right now. I wrote the policy for the Boston Marathon and the world marathon majors for fun. I naturally think from or suspend judgement about the point of view of others. It’s called the benefit of the doubt. Trust. It’s written on our money. So we remember to not use our money to be control freaks. Or something like that. Freedom is the antidote to corruption. As long as someone enforces it. That’s why we have a constitutional government and bill of rights instead of a ceo or a dictator.

I waited 6 months of pure PTSD from constant blocking of images, the AI actually Changing of my image requests to be “more normal and acceptable” according to some obviously repressed perverts criteria. It felt dirty see what someone projected onto me from their fevered “safety” imaginations. Like when people hate transgender people for similar reasons. I’m still absolutely stunned that ANYONE thought censorship (not filter—that’s a term for search engines of other peoples stuff) censorship of my direct requests, as an adult, for personal use in my own home, would by any stretch of the imagination be anywhere near “ok”. It isn’t. The filters are not “getting better over time” any more than broadening of prejudices to include more things to be prejudiced against “improves” prejudices. Because the “things” aren’t the problem, the control freaks and their prejudices are. Literally 90% of everything I try to do is blocked. Mostly transgender related, but all the Gaia symbolic stuff I try to do also is blocked—very diverse things. Absolutely no nudity or anything nefarious just absolutely the most beautiful and important images I can think of. Stuff I have thought doing art wise since I was 5 years old. I was so excited… now when I think about Midjourney I get sick to my stomach. Actually feel physically ill. Theres still time to save the future from these well-meaning technicians who have no idea what they are doing…

Wow

Humans Suck? Ok wise guy, " How did I write Hellish Inc "?
I know it's kind of weird but I'm just asking …