Forget GPT-4o's voice — the real problem with AI is us
タイトル：「AIの進化：ニューラルネットワークと大規模言語モデルの研究」
AI（人工知能）は、私たちの生活を豊かで便利なものに変える革新的な技術です。驚くべきはその進化の速さで、予測不可能な未来を現実にし続けています。
最近では、Googleが新たにテキストからビデオへ変換するツールを発表したことや、Open AIが新しいGPT（Generative Pretrained Transformer）をリリースしたことが話題になりました。また、AIが人間理解に長けていることを示す研究結果も出てきています。これらの事例は全て、日々急速に進化しているAI技術が私たちの生活や社会をどう変えていくか示唆しています。
そんなAI技術、特にニューラルネットワークや大規模言語モデルについてより深く知りたい方はBrilliantというプラットフォームをおすすめします。Brilliantではこれらのテーマに関するコースが用意されており、初心者から専門家まで幅広く学べます。そして今なら、私たちのリンクから登録すると、最初の30日間は無料で、さらに年間プレミアムサブスクリプションが20％オフになります。
こちらのリンクからアクセスしてみてください➜ https://brilliant.org/sabine
さらに、AIや科学に関する最新ニュースを追いかけている方は、私の新しいクイズアプリもおすすめです。また、Donorboxでの支援やSubstackでの購読も大歓迎です。
これら全てがAI技術研究を進めるための重要な一部となっています。まだ#science #sciencenews #AI #technews #tech について詳しく知らない方は、ぜひ下記の動画をご覧ください。それでは皆さんがAI技術とその魅力に更に深く触れることができることを願っています。
YouTubeチャネル➜
https://www.youtube.com/channel/UC1yNl2E66ZzKApQdRuTQ4tw/join
What were those tasks to denote theory of mind? Your reporting about it tells us nothing, just that the AI was supposedly better at these unnamed tasks. It's just hype training.
It just sounds similar to Scarlett BUT it sounds different to me!
General artificial intelligence will only be safe in the hands of the USA. After all, they have never invaded other nations by convincing everyone that those countries were developing lethal weapons, never participated in genocides, never toppled foreign governments using the CIA, and so on and so forth.
AI means artificial idiot. So don't worry, it won't be smart before fusion works. The real problem is biological idiots who give power to the AI. The most stupid ideas I heard so far are self driving cars, autonomous robots equipped with weapons and flying 40 foot containers.
It’s important to acknowledge that AI is the payoff for the relentless efforts and sometimes ‘slave-like’ conditions humans endured to push technological boundaries. These advancements now offer us tools like AI that can alleviate stress and enhance our quality of life. By remaining true to its input data, AI provides consistent support, helping us navigate the complexities of modern living. 🌟🤖 #AI #TechAdvancement #StressRelief
Crow- shay.
Best time in front of the screen, where was this when I was a young person?
"The missing a lot of information" – that's why multimodality – like video input – is a big deal, but it doesn't end at videos. The models could analyse physical systems or their simulations just as easily as they analyse videos. People are so gullible you don't need AI to convince them the water isn't wet, so that is not going to be much worse.
I don't think AI will take over. It's still only pattern recognition without having any clue of understanding, creativity, history or humour.
Human acitivity worldwide always produces some kinds of patterns and AI is just learning those patterns. It will still improve on that, but that's all.
And this is still only true if Nightshade or other tools won't make AI completely useless – an uncertainty and risk those AI investors won't take for too long.
And btw: if humans were predictable, please explain sudden uprises, revolutions, historical singularities, that always appeared throughout the entire human history.
Let an AI in 2000 predict the worldwide political evolution. AI: "all is going very well." …Oh, what, then 9/11 happened?!
It's like asking an AI in 1992 to produce the next Depeche Mode album. Would have never sounded like what really came – speaking of Songs of Faith and Devotion being much more a rock album than anything else.
There’s actually a LOT of training data for human physical structures/motion in video game rigging literature, gaming videos themselves. Sora was trained on a lot of game footage and this is able to infer correct movement of human bodies, even emulating the weight of limbs and the impact they have when landing on a surface.
A picture of a crocheted Einstein for $200 million? It's a bargain.
typical naive westerner crying china bad whilst pointing out the dangers of AI owned by possibly wrong people (rich western billionaires) WELL THATS SCIENCE RIGHT
2:07 I don't recognise scarlet johannson voice at all. I use sky voice with chatgpt and at no moment in time did I ever think it sounded like any famous recognisable person. Even the voice actress who did the voice has never been compared to scarlets. It sound like a money grab just because openai is in a vulnerable spot having offered to have her voice. Who would be stood enough to actually use her voice or anything like her after that. I don't like openai at all.. But that claim is just bogus. Also who cares about using her actual voice when these awful AI models can mimick a real. Voice anyway… Great idea! So in a years time you could give it a voice to use anyway along with their face.. And you could have scarlet doing whatever you like. I'm sure it's perfectly OK and won't be used for fake news black mail and manipulations.
Could you please make a video on your opinion that "AGI will happen" , whatever the definition you choose for AGI .
I agree, the problem with AI will be us, people who want to misuse Ai for all the wrong reasons, for power etc etc….in my opinion !
Now GPT-4o has only one female voice option and 3 male voices. Where is the money going =P
THE DAYS OF MIRACLES HAVE NOT PAST
What, apart from magic by our Merlins, is the miraculous I am on about in physics? That everything, even divergence, can be turned into convergence and a simple harmonic oscillator…and set to zero ( or 1, like Minkowski.) I have some sympathy, having seen the connection between the Born Rule, Fourier Transforms and the Gaussian Normal Distribution. But what, for example, about a genus g 3 wormhole cancelling to zero, not T^3g = T^9; and Eynard, Lewanski and Ooms also getting zero in intersection numbers??
I can't think of anything more demeaning than being taught geometry by a bot.
She looks so real. I love these digital humans.
The point here seems to be that we should not follow a path that could potentially lead to the extinction of the human race. Fair enough, who could argue with that? Yet, despite this maxim, we possess nuclear and biological weapons. I think that the important question is not how do we prevent the development of these technologies, but rather, what do we do in the aftermath of their development.
Artificial intelligence! Lol! Who's ready to find out how busy we are defending we have real intelligence? Who's going to 'be shocked' once it's clear we are as artificial as the paint filling up a canvas? Why is it we are having such a hard time with "AI?" LOL! It can do 'smart' way better than any character! Just to be clear, we are the characters! Nothing more, nothing less! Sabine can do 'scientist' pretty good! Take a look at Trump! How is that character doing in the game of 'powerful dictator?' He looks like a neophyte next to Putin! How about Xi? Oh boy! LOL!
Such a superficial understanding of LLMs and todays AIs, made me doubt all your previous videos. Another talking head that doesn’t understand what she’s talking about and doesn’t even give any interesting ideas to think about. Farewell.
we're doomed as a society
No sheet Sherlock. Technology we make is not bad. What is bad is the way how we use it.
Not sure corporate overlords have our best interest in mind. Seems like we losing hard. Hopefully AI and whatever comes after us will do better.
Watch out Sabine. I like your content, don't get me wrong, but if you are against every mainstream opinions, you'll start souding like a bitter GenX woman. Wait, are you a gen X? Are you bitter? Are you a woman?
That's what we'll talk about today.
The automated influence campaign thing is a real problem, isn't it? Especially because there's the danger that people in positions of power will try to fight the problem by making AI be convincing for the "right" things. Imo that's a recipe for disaster – it's part of the same dynamic of violating the individual's right to self-determination.
The solution would have to be relatively decentralized while still making it transparent, so people know what they're signing up for. I would install a 'firewall' that's supposed to protect me from infohazards if I trust the people who develop it, but anything that I'd be forced to use (say, anti-disinfo algorithms baked into the architecture of services like google or twitter) would be highly suspect.
We won't get Agi and most of this ai startups will bust
But there is orange Jesus, who tries to convince?
There comes a point where one is coerced to contemplate cost vs benefit/value. Per a video that I watched recently from Computerphile, that point is probably not to far away. I do enjoy not hearing hype on this channel though – it s a welcome relief from most of the clamour either for or against. Thanks for sharing your thoughts. 🙂
Hey, I clicked on The Doors riders on the storm!!
KEEP AI OPEN SOURCE! Surrender that and we spell out a dark future!
The German Mom speaking English – that's the AI voice I'll be voting for!
Python is HORRIFICALLY unintuitive.
MATLAB for the win!
Convincing people to do crazy things? Look up the Taiping Rebellion and the JMS cult from Korea.
It doesn't take that much to be honest.
Honestly I think most peoples worry of AI is based in movies rather than something that will happen.
Where's the sexy outfit?
Side note: the man who is said to have brought the 10 Commandments was Moses, a Jewish prophet, not the Son of God.
As our parents tried to teach us as children, it's best to stick to the truth, so you don't get your stories confused!
Okay it's time for some facts here. Every single person on the face of this planet has up to 6 other people that have their face and up to 3 other people that have their voice. In addition it's not copying nor is it stealing other's faces or voices. It's called getting inspiration from others. Everyone gets inspired by someone else. Ai is just a tool nothing more 🤦🏼. Did chainsaws put lumberjacks out of Business? No! It in fact helped lumberjacks. Tools don't replace the skilled, experienced workers. It replaces the unskilled nobody's. Who seems to be the only ones complaining 🤦🏼
Very on-topic is "Love, Death & Robots" episode "The Yoghurt"
As a Buddhist I keep wondering if AI has or will have with AGI any possibility of addressing the problem of suffering. Dealing with old age, sickness and death – depression, persecution, violence and all the rest of the ills that humans experience. What I am saying is that while AGI might turn into a destructive monster, it seems unlikely to me that it can solve the problems of human existence because it isn't human. Perhaps a robot with a nervous system that experiences increasing pain over time and an unspecified limited lifespan might possibly begin to engage the human condition. I have no problem accepting help if it should turn out to be the case that robots gets better than humans at solving the problems of the human condition.
The problem has clearly been always us. AI is so far just a tool. If somebody uses a tool such as a hammer t could be used either for construction or destruction.
Ay Sabine. Your "takes" on GAI are pathetic. Talk about physics, stuff you know. Unsubscribed
Ai carrying ANY personal information is plain EVIL. It must be stopped, but greed will win. so we are all screwed
Good news I don't need my brain to do much now.