Thanks to MANSCAPED for sponsoring today’s video. Get 20% off + free international shipping with promo code TECHTIPS or visit Explore MSI Gaming Desktop PC in AI is the buzzword de jour and for good reason: it enables MANY new and useful tools for everyday life. But what exactly IS? And why do we think it’s unfairly represented? Discuss in the forum: Purchases made through some store links may provide some compensation to Linus Media Group. ► GET MERCH: ► GET FLOATPLANE EXCLUSIVE CONTENT: ► GET VPN: ► SPONSORS, PARTNERS & PARTNERS: ► EQUIPMENT WE USE TO FILM LTT: ► OUR WAN PODCAST EQUIPMENT: FOLLOW US ———- ——————————————— Twitter:.. .

By admin

25 thoughts on “AI is a Lie – Cutting Through the Hype”
  1. If your definition of ANI is that it cannot by definition handle a new edge case, then no the most advanced models of today are far outside the definition of ANI.

  2. AI is not a lie, what they claim about AI are mostly lies, deception and exaggerations; and this is what you are talking about indeed but it seems you need a flash title as usual. I tell you what is an obvious lie is though… AGI (Artificial general intelligence).

  3. At what point do you become guilty of misleading individuals when you name your video "AI is a Lie" -a general proposition, which is, on its own, patently false. It seems to me that if you're making a value judgement predicated on honesty, you have to be honest yourself.
    Ancillary to that, human individuals have a "narrow slice" of intelligence based on their own "training data", and are also capable of every single one of the errors you mentioned; so unless you're willing to concede that human intelligence is not a true form of intelligence and that you are additionally advancing the claim that your own honesty is exempt from the conversation, your commentary cannot be received as credible.

  4. Quite an interesting and informative video that will probably get a re-watch by me. That ELIZA clip – what a blast from the past. Many moons ago I had an interest in AI [not what is called AI via 'buzzword' these days] and ANI – at the time that meant Artificial Non Intelligence. I had a version of ELIZA installed on what was probably my first desktop PC and enjoyed playing with it. I also had access to a version of NIAL! Too many memories from way back when. I'm still very interested in all things 'computers' and LTT is, IMHO, a great channel for my 'tech fix'. Keep up the good works.

    Edit: Anyone seen the crap where a company uses 'Automated' instead of 'Artificial' – as in 'Automated Intelligence', BTW?

  5. Watched a video from this channel after a very long time, and I was not disappointed by the complete and utter ignorance about the subject.

    GPT-4o and Claude 3.5 Sonnet are definitely more intelligent than anyone working in LTT…

    I've been using LLMs since GPT-2, and I have more than 4000 hours of use in multiple fields (from coding, to game design, brainstorming, and writing), and also a few hundred hours with image and audio/music generation models.
    At this point, Claude 3.5 Sonnet is much better at creating entire apps than developers with even 5-6 years of experience (not even talking about time and cost).

    If anything, AI is underhyped. The only negative, is that a lot of companies are hyping it for the wrong reasons, and that's because most people are completely ignorant.
    And most people crying about AI being hyped or talking about lies (not everyone is Google, and their performance and abilities are an undeniable fact at this point) and similar BS, not only have barely any experience with them, but they have definitely never read even a single research paper related to this field (and there are hundreds coming out every month with insane breakthroughs all the time).

    edit: Also, about Tesla's FSD. Do at least some simple research next time.
    FSD v12.5 is better than the average human at this point, and of course much safer.

    Anyone that thinks AI is a lie and overhyped (in other words completely ignorant about the subject), is free to come again in 3 years from now to tell me if I was wrong, I will be waiting.

  6. Ai is a broad term, it never meant the same thing as AGI or meant conscious, self-aware, thinking machines. That was never the original meaning of Ai, that was the hollywood movie definition of Ai. Claude Shannon's Theseus, an electromechanical maze-solving system is Ai. Pac Man ghosts with their 3 if statements is Ai. Ai can be really simple and basic, people watching movies define the term Ai in context of these thinking talking sentient characters. That can be considered Ai too, but extremely advanced Ai, that does not mean that simplistic Ai is no longer Ai. LLMs are Ai, it is not intelligent, it can't think or feel or understand, but it is still Ai. Blinky, Pinky, Inky and Clyde can not think or feel, they do not contemplate their existence or which way to turn to get Pac man, they just turn left right or go straight all depending on where Pac man is on the map. That is still Ai. Ai can be simple and dumb. People think Ai is R2D2, Chappie or Johnny 5, one thing they all have in common is they are all movie characters! People are taking the movies too seriously, they apply what they see in the movies to the Ai marketing going on right now. Ai is not a lie, all the stuff that claims to be Ai is Ai, the real lie is movies, fictional movies. Fiction means not real, so people should not be applying how terms are used in movies to real life. It's that simple, what most people consider Ai to be is the lie. Ai can be simple, or it could be complex! It could be a walking talking self aware conscious being, but that doesn't change that Theseus is also Ai. It's a broad term that encompasses a lot, that's why people say things like hard Ai and soft Ai or AGI, to differentiate the different kinds and levels of Ai.

  7. I’ve been playing computer wargames since 1981 and I can assure you the AI of today is not substantively better than what we had in 81. The only “good” opponents use scripting and canned scenarios (which kills any sort of replay-ability).

  8. I think Tesla might have solved FSD just needs to put the stacks together. You are correct that sooner or later an edge case will come up that the llm can't handle but overall it's much safer than average human driver. Otherwise your video has aged very well.

Leave a Reply

Your email address will not be published. Required fields are marked *