Comments by "ΞΞΞ" (@isfrank) on "AI Search" channel.

  1. Yeahh, I'm not submitting any personal information to a Chinese website.
    57
  2. I would for example absolutely love to work a few hours a week and work on big creative projects in my free time.
    25
  3. "AI is more creative than humans" Yeah? Ask it to come up with a new color. I'll wait.
    7
  4. You forgot to copy the js line for that coffee html page... granted, you prompted to give everything in one output and was in a seperate codeblock, but it was right beneath it
    5
  5. I ran the 70b model locally just a few days ago and it hallucinated a lot, so I uninstalled it after just 3 prompts. Their top of the line version is a lot better, but it is still far from the competition. Gemini will soon catch up. That side-by-side will be fun
    4
  6. Creativity. Intelligence is true creativity.
    3
  7. Women: left, but both eyes much too red Soldiers: left, but far from realistic physics Zombies: left, but far from good enough Children: left, but hands Will smith: ...left? But this is becoming weird now Ship: none, they failed the prompt Panda: none, they failed the prompt Dragon: none, they failed the prompt Anime: none, they failed the prompt Brush: nah, that aint it either. Conclusion: hollywood? Far from it.
    3
  8. If the data is human, the result is human. The answer, based on that, is probably yes. Don't blame AI, blame the human race.
    2
  9. The real test is crativity. Your prompt falls in the category of how thourough or recent the crawled data is and how well it can replicate what it already saw. A real AI can take info of a million things and create something truely unique. Intelligence is very much aligned with creativity. Never forget AI stands for "Artificial Intelligence", not "Fast photocopier with a gigantic database"
    2
  10. Try to prompt the big LLMs with this: create a 3D-style ASCII art representation of "LM" It will always fail, for some reason. Either it gets the characters good, but no 3d, or just a completely wrong abstract result is returned... depending on the llm It doesnt require the letters "LM", you can use any but I think it starts going wrong at 2 or more the most
    1
  11. Addition: I can confirm now that the bigger the set of characters, the more wrong the answer gets
    1
  12. The reason tge covid vaccine was developed quickly was because they were allowed to go ahead and continue development WHILE certain checks or certificates (not sure how they're called) were still pending. This significantly sped up the process and is the main reason the vaccine reached market that much sooner. Another reason is the fact that it's mRna, that helps too.
    1
  13. The game; so an AI, with agents that had at least some previous training (probably on human data), developed traits exactly like humans.. what a shock! -_-
    1
  14. The statement "we can only create new ideas from things we've come across" (or similar) is false. To find examples, search for unique children's creativity
    1
  15. The proteine binding, is that successrate established through practical studies, or theoretical?
    1
  16.  @gdok6088  cool, thank you
    1
  17. I'm just a man, so all I can do is connect dots. If an AI gets data by crawling the web And it can generate images based on what it crawled When it generates 3 our of 4 images that violated content policies That means our internet is at least ¾ porn.
    1
  18. ...or some very horny crawlers, either must be true
    1
  19. I asked chatgpt4o: "How many r's are in Strawberry?" It responded with: "There are three "r"s in "Strawberry.""
    1
  20. As I predicted ;)
    1
  21. Can we AI-generate blender models yet?
    1
  22. I don't know if you are in the USA and it is just configured to be dumber for people from there, but the strawberry question response for me is always 3 r's
    1
  23. Dell and Nvidia sponsored you? That's amazing, m8. It's not like you have millions of subs (yet), so yeah.. amazing :) good for you
    1
  24. You've again sparked my curiousity for a new model, but last time (Llama) after the first prompt, I already knew the model was terrible. It took only 4 more prompts for me to know I will not use Llama for the forseeable future. Good luck, Glaive!
    1
  25.  @remora8829  I'm well aware, you seem to not understand that I would rather have my neighbor manage all my data opposed to someone who's actions I can't predict as a consumer.
    1
  26. "AI", for me, means a software driven intelligent entity, just as a human is an intelligent entity. Any reasonably mentally well developed adult can, at all times, respond to whatever it is presented / audibly prompted with, in [㎳] (mili seconds). Even behaviour and sounds of animals (limited average to the species with globally most human interactions). If my goal is to render you, a human, genuinely speechless, I would have a very hard time. For this comparison, let's view severe hallucinations as speechless for the latest LLMs; I would have no problem with confusing GTP4o. Not textually, audibly, nor visually. As soon as it can respond in realtime to how I talk and which social queues I present, either through body movement, or lack there of, or tone in voice rhythm, choice of words, length of pauzes... basically everything that defines human communication, only then will I accept it being an Artificial Intelligence. Right now very smart people programmed something and that something does what it has been ordered to do. It can't really develop "mentally". Scope is the limit. It is another very advanced Automatron. Not intelligent. It has no character, it is not unique for you or me. It does the same for everybody. Automatron.
    1
  27. Software version 9.11 is actually higher than 9.9 Also, if the text On a screen is Bunched together Like that, it doesn't automatically mean that the answer is longer ;)
    1
  28.  @SixTough  yeah, because in tech (versioning) the digits are usually seperated for Major versions, Minor versions and Patches (1.2.3 respectively). The reason why 1.5.0 lower is than 1.25.0, is because versions are incremented sequentially and only return to 0 if the leading number increments. Meaning: 1.2 becomes 1.3, 1.4, 1.5, 2.0. Back to 0, because the new Major version has yet to receive any Minor updates. There is no subtracting or dividing in versioning, because it's not Math. Everybody (usually, roughly) knows what happened before and what can happen next without having to calculate variables. Returning to the AI topic; If the context (i.e. topic conversation or the user's possible profile weights) is not software related, in my opinion the AI should answer a question with real world applicable logic. Since in every day life we deal with many types of numbers, of which time the most, assuming 1.25 is less than 1.5 would be the logical approach. Possibly with a small (*) that mentions the reasoning
    1
  29. Seeing as dell and nvidia sponsored this video, did they gave you those prompts to use also? Or maybe a list of prompts for you to choose from?
    1
  30. Google Learn About is potentially the new Wikipedia
    1
  31. Regarding the math problem; Was it able to reference the answer somewhere? Was it cached? Is the math problem posted publicly? If yes, posted publicly how long before the time of query?
    1
  32. "Can't be smaller than this" is not true. 0,4nm is scheduled for release by 2035
    1
  33. Here I am again, making sure people know this is not Artificial Intelligence, because intelligence directly correlates with creativity. If something acts on what it is trained with, it may react 'smart', by being fast and accurate to the input, but it is not intelligent.
    1
  34. Programming is, at best, as smart as the people who made it.
    1
  35. No thanks. I'll stick to the MVP; Anthropic's Claude.
    1
  36. If only you did your research before making certain statements 😂
    1
  37. There is no cake #oldreference
    1
  38.  @spookaygod7782  that's a hl ref
    1
  39.  @Alice_Fumo  that's the idea of a color. I want it to show me this new color. It can't. We can't.... usually. Sometimes it happens, but after adding a few more colors to the rainbow and then some, we kinda hit the limit of our eyes.
    1
  40.  @UlyssesDrax  look, I'm being serious. AI is a serious topic. If you want to be like that, maybe do it somewhere else, alright?
    1
  41.  @UlyssesDrax  well I am very sorry if I didn't formulate perfectly on youtube in my third language. Very sorry for your inconvenience. I will try to do better in the future
    1