Views : 165,849
Genre: Education
Date of upload: Sep 1, 2023 ^^
Rating : 4.873 (257/7,861 LTDR)
RYD date created : 2024-05-20T15:54:15.58982Z
See in json
Top Comments of this video!! :3
In my modest opinion as a big fan of Ready Player One, I LOVE the concept behind it, but I just think Zucc did the worst possible thing - set on fire so much money and had to show for it something that looked like 3 people in a garage with a potato pc and 3000 dollars would make. Maybe starting smaller, making something less ambitious but incredibly good might have saved Meta a shit ton of money while still creating something truly impressive. I know one day a big brain gigachad will come and create The Oasis, but till then, weâll keep getting disappointed, ig. Wonât even get to see the Zucc vs Musk fight, that was another letdown from our reptil friend
188 |
I worked for a tech giant for a very long time, and Product was so slow and bureaucratic that small companies with 1/10th the development team were able to code circles around them and overtake them. I watched with amazement as a startup competitor came out with in-demand feature after feature while our Product team coded things no one asked for or wanted, and took eons to do it. I feel like Musk has somewhat cracked the code for avoiding this by slashing Twitter down significantly and orienting it like a startup with a new mission. But most big tech companies canât do that. They should play supporting roles in other tech, but once they start trying to get directly involved in something else they usually fail. They also have to keep their Product team fresh and avoid bloat. Bloat can be useless developers or simply useless job roles. All bloat adds time and cost to innovation.
49 |
Great video, as always.
I will say though, Google is MUCH MORE risk-averse when it comes to AI.
But make no mistake, theyâre ABSOLUTELY a leader in AI research and capability.
I meanâŚthey LITERALLY CREATED the Transformer at Google Labs.
In fact, by releasing the paper âAttention is all you needâ and gave it out for free, they effectively enabled EVERY SINGLE STARTUP that utilizes LLMs today.
Google Bard is shitâŚno doubt about it.
But itâs nowhere even close to being powered by their most powerful model.
Not to mention, they are set to release Gemini, the first LLM built from scratch for multi-modality in the next few months.
..and Gemini 2 is already reportedly in training.
I guess my point isâŚno one can sleep on the literal creator of the technology that is causing all this mass hysteria.
65 |
The reason I never got into Facebook is that it replaced a platform that could be customized into uniquely individual spaces with a branded uniformity, so that nothing appearing in THEIR space could ever reach highs or lows, and any 'personal' space encountered would never be that personal. It's lasted much longer than I expected, yet I still don't care to give it any of my effort or attention.
32 |
Google may have worked on AI first, but they failed to be first to the market, and they flopped very badly after rushing to get it out to the market. And worse yet, they proved everyone wrong who claimed that Google was years ahead of the competition and just didn't want to release it yet because it might be a threat to Google Search.
15 |
Your right the titans are throwing a lot of money at it, for diminishing results. However for companies that collect data, using ai is very profitable because you now are able to figure out what the person is thinking about much clearer. Imagine if someone using chat gpt on writing a report and the data is then processed by Microsoft for providing ads to that person when they use edge. Or if it's research related companies would like to hire people who do that type of research like pharmaceuticals so they can send job ads to those who use ai assistant for specific jobs.
38 |
The way AI is structured, i.e. make bots, train bots, promote/demote bots after a test, rinse & repeat - is very limiting.
People recognise that the army of bots which are assigned to solving a task - operate following a black-box model, and that is true. The black box hides the topology of the network where a sequence of logic decisions bring to the final result; this structure is very limiting, because you can correct any error exclusively by trial and error.
This has never happened in the history of human engineering; sooner or later, engineers first devised an empirical method to calculate a machine or device, and successively they always found an analytical methodology for locating the proper solutions.
Arrays and matrices of bots don't scale up very well; the telephone engineers of 1920 faced the same problem when the telephone network expanded first locally, and then internationally. The future of AI will require that software engineers join forces with mathematicians, and together they intervene inside the black boxes.
Thank you Harry for this video on AI. In parallel to what you said in the video, I like to add that the AI development is akin to the Race of the Red Queen in Alice in Wonderland...
9 |
@tombouie
8 months ago
The next big-thing is regulating those darn mega-tech companies just like any other darn utility
587 |