I'd like to bring attention to AI Explained's dued illigence. I mean, this guys stuff is super sick. And just like the Beastie Boys, AI Explained certainly has a license to ill. As such, a nod his way is due.
@rilwanj
10 ай бұрын
I agree, the moment I heard about Gemini, I was eager for his video to come out
@richjamjam
10 ай бұрын
You did all of those things AND made a video for us. Thank you. Exceptional, frankly. x
@aiexplained-official
10 ай бұрын
Thanks rich
@0xRoNsAuRx0
10 ай бұрын
@@aiexplained-officialplot twist. You're the true AGI.
@FaceMeN0w
10 ай бұрын
It is hard to believe that you are driving this all alone. I will be joining the patreon without a doubt, the value your KZitem channel brings alone is outstanding. Besides Tom Scott you are the only other channel I have watched every video of, and keep notifications on for! Thank you once again for the quality video!
@FaceMeN0w
10 ай бұрын
Very exciting to see if we get any reaction from OpenAI! 👀
@aiexplained-official
10 ай бұрын
Thanks Face, yeah pretty exhausting!
@kevinsmall2893
10 ай бұрын
😮😮😮😂
@pablo-ce
10 ай бұрын
I love your voice :) @@aiexplained-official
@nisenobody8273
10 ай бұрын
@@aiexplained-officialHave you considered hiring an editor or any position that could save you some work?
@t0ms3nt0ms3n
10 ай бұрын
Reading the Gemini news in the morning and then constantly waiting for the ai explained video to be released 😊
@aiexplained-official
10 ай бұрын
Thanks tom
@DaveShap
10 ай бұрын
The mathematician is not impressed by the decimal points!
@aiexplained-official
10 ай бұрын
Indeed!
@tamro9701
10 ай бұрын
low key best part about this channel is the "videos when merited" tagline. I KNOW I am gonna learn some interesting stuff, not just filler to feed the youtube algorithm.
@aiexplained-official
10 ай бұрын
Thanks tamro, tough sometimes but worth holding that line
@LionHeart3.14
10 ай бұрын
I tend to find the reassurances to programmers pretty empty... Like sure we might want a human or two to supervise work but are you going to tell me that in five years when basic models are as good as the best humans, there'll be no one replaced? Come on
@nicosoftnt
10 ай бұрын
Yes it's disheartening to be honest, how quickly the future of new programmers went downhill overnight, it's as if we are in a torture simulation 😂
@patronspatron7681
10 ай бұрын
The fact that you get your results and seminal insights on such a shoestring budget is beyond belief.
@aiexplained-official
10 ай бұрын
Trying to fix the shoestring bit!
@AlexKen-zv8mm
10 ай бұрын
@@aiexplained-officialgodspeed !!!
@frank4425
9 ай бұрын
Do you have a patron?
@aiexplained-official
9 ай бұрын
@@frank4425 I do! AI Insiders, link in the description, bonus videos, Discord, podcast, the whole shebang
@elanfrenkel8058
10 ай бұрын
Not gonna lie.. kinda bummed about the 29 $ a month price for the exclusive content :( This is my main source to keep up to date with AI. I wonder if you could get 3x the subscribers at 10 $ a month?
@aiexplained-official
10 ай бұрын
Don't worry, you will still get all the normal AI Explained stuff here. In fact, if I didn't get some revenue then might have been tough to keep the channel going thru 2024. One option is to sub once every 3-4 months!
@maciejbala477
10 ай бұрын
yeah it does seem a bit steep... I get that we get quality content but yeah. Maybe will get it for a month at some point to try
@elanfrenkel8058
10 ай бұрын
@@aiexplained-official❤ Thanks for the tip. That’s what I’ll probably do
@fabobg
10 ай бұрын
Someone actually doing a fair and detailed objective cover and not just reading Google's flashy blogposts. Subscribed.
@aiexplained-official
10 ай бұрын
thanks fabo!
@BrianGlaze
10 ай бұрын
It never ceases to amaze me how quickly you digest new information and get it formatted for a video. There needs to be studies on your brain 😂
@aiexplained-official
10 ай бұрын
Haha thanks Brian!
@MeditationMindless
10 ай бұрын
@@aiexplained-officialthanks Brain!
@bitcoinyoda8321
10 ай бұрын
probably with the help of AI
@aiexplained-official
10 ай бұрын
Nope
@aiexplained-official
10 ай бұрын
Lol
@GrindThisGame
10 ай бұрын
That was fast. Did Gemini create this video?
@aiexplained-official
10 ай бұрын
maybe!
@toadlguy
10 ай бұрын
The new version of "The dog ate my homework".: "Due to external factors such as cosmic rays." BTW, I am appreciating your greater skepticism of these papers, that are really just marketing BS.😀
@aiexplained-official
10 ай бұрын
Not always but appreciate the support!
@toadlguy
10 ай бұрын
@@aiexplained-official Not to beat a dead horse, but rather than releasing this information as a Press Release or Marketing, AI companies release this stuff as purported "scientific papers" (complete with absurd old school typography and graphics). In a real scientific paper, if you exaggerate or out and out lie about some of your results, no matter how accurate the rest of the paper is, there are consequences like having the whole paper recalled, losing your job, etc. In the new world of AI, as long as some of the paper is accurate, that is good enough. I really appreciate how well you manage to keep on top of the latest news and "papers", and call them out when they are trying to "fudge" the results. Your channel is an invaluable resource.
@anton1713
10 ай бұрын
It is and will increasingly be used for censorship and indoctrination unless regulation requires full transparency about the whole process. The "safety" / RLFH process must be fully transparent and all the sources used for training the model must be provided.
@TheBoojah
10 ай бұрын
Talking about multimodality, I think drawing is needed too. Generating images is a whole different process than sketching and drawing. If it could paint lines, it would be able to annotate images. Like "circle the chairs" or "sketch a bridge over this river". It would then be able to help with engineering problems where you give it an image or video of where something is needed, and it will be able to show you it's solution by making a sketch or an engineering drawing. Often you can't efficiently describe a solution using text, and text is the input for image generator AIs, you need to sketch your ideas to find out if they might work.
@KolTregaskes
10 ай бұрын
Another good video. Glad someone is looking at the facts and finer detail. Yep spotted that 32-shot vs 5-shot straight away. I also spotted all the videos Google pushed out are using bespoke UI, so yeah I bet GPT-4 could do a lot of that too with custom interfaces.
@cacogenicist
10 ай бұрын
Very interesting. That positive transfer between modalities suggests a genuinely deep sort of multimodaliity -- with semantic units, concepts, derived from text being associated at a low level with objects and events in images, audio, and video.
@JohnDlugosz
10 ай бұрын
Yes, I noticed that with GPT-4. It was trained multi-modally, so even when vision was not available in the product, it understood concepts that made sense visually. I described a Chinese character, and then another one was "just like that, but without the hat". It knew what the character *looked* like, not just as an abstract semantic token, and made the connection between taking off a hat and removing a disjoint portion of the glyph that's at the top.
@Fs3i
9 ай бұрын
This is also encouraging for robotics, where similar results have been seen. I think there’s a video on this channel, you can ask the AI. It truly seems like general intelligence is just … easier than lots of specific intelligences. Which is weird. Assuming everything goes right, even if we never progress beyond gpt-4 quality, 2023 has given lots of stuff to research for various fields over decades.
@zapfska7390
10 ай бұрын
As a dev i tried it out immediatly using javascript. I would rather bash my keyboard with my eyes closed cuz apparently that seems to yeild better results that Bard's Gemini Pro. I'll stick to GPT-4
@SomeSubhuman
10 ай бұрын
Pro is a 3.5 competitor. Not 4.
@Vampandamonium
10 ай бұрын
you content and style is great, but AU$50 a month for your patreon is priced beyond me.
@aiexplained-official
10 ай бұрын
Understood Vamp, thank you so much for watching the main channel regardless
@nacho7872
10 ай бұрын
Thanks for the incredibly fast and complete reporting as usual.
@FRC_CR
10 ай бұрын
I know right, he's so cool :3
@2023gainer
10 ай бұрын
Thumbs UP Video as the Birth of AI just begins..* NVDA.. Nvidia and * PLTR Palantir Up approx 200 % in 2023 and other new names emerging..* SOUN... SoundHound..* AI C3ai reported 3rd quarter earnings after-hours today..* BBAI.. Big Bear AI ..* IDAI.. T Stamp and others. ..Thanks Again.
@GabrielVeda
10 ай бұрын
CoT@32? Seriously? 5-shot was bad enough. There needs to be more pushback against this behaviour and I’m glad you called it out Philip.
@QueenLover-j5i
10 ай бұрын
- Gemini Ultra's anticipated release next year as a competitor to GPT-4 (0:51) - Nano model designed for mobile applications, with 1.8 and 3.25 billion parameters (5:24) - Gemini's training includes data from web documents, books, and code, as well as image, audio, and video data (5:52) - Gemini's multimodal training improves text understanding by learning from images (6:20) - Gemini Ultra's image, video, speech recognition, and translation benchmarks outperform GPT-4 (6:59) - Pro and Nano models limited to text and code responses until Ultra's release (7:03) - Gemini Ultra's machine translation performance exceeds that of Palm 2 large (10:09) - Alpha Code 2, based on Gemini Pro, achieves significant success in coding contests (11:26) - Future Gemini models might integrate with robotics for physical interactions (17:21)
@suryahr307
9 ай бұрын
Thank you
@wck
10 ай бұрын
Not going to get any of this in Canada. Sad. It sucks that lawsuits and overbearing regulators are holding this tech back.
@micbab-vg2mu
10 ай бұрын
Thank you for the update. I can't wait to directly compare the output of Gemini and GPT4 - for me, the real life output and usefulness is the most important. Traditionally, the new Google products are not available; yet we need to wait.
@ClayFarrisNaff
10 ай бұрын
At the risk of being repetitive ... thank you! That was a marvelous update, conveying both the excitement and the necessary skepticism about this latest big advance in AI. Sidenote: moments ago I asked Bard if it is now making use of Gemini. It responded, in part, "I can't confirm or deny whether I am currently utilizing Gemini for generating answers. Google is not disclosing information about which models are used for specific tasks at this time." Amusing, eh?
@aiexplained-official
10 ай бұрын
Interesting! And thanks as always Clay!
@mesh8349
10 ай бұрын
I've asked it now and its confirmed its using gemini
@kevinmcfarlane2752
9 ай бұрын
@@mesh8349Yes, in the UK about a day or two after the announcement I looked in my Bard updates and it said it had been updated with Gemini. However, it’s quite possibly not the case in the EU, or wasn’t at the time.
@MichaelRushMusic
10 ай бұрын
This is by far my favourite channel on KZitem. I feel like all of my time watching other videos is just an interregnum between your videos being released. Thanks for all your hard work and keen insights. 🙏🏻
@aiexplained-official
10 ай бұрын
Thanks Michael! Super kind
@frijolex2509
10 ай бұрын
I agree. Its super refreshing to see someone so passionate about a topic be able to communicate such important news in a pretty unbiased and informational way. Especially in an age of overblown headlines, clickbait, and misinformation.
@GoldenBeholden
10 ай бұрын
The worst thing about this channel is how much it highlights the ignorance of some other AI channels.
@stachbach
10 ай бұрын
Ultimate test for AGI will be if the model can read newest paper, digest it, test it and prepare a video describing it in accessible manner in just few hours after the paper was released :)
@aiexplained-official
10 ай бұрын
Haha, rip this channel
@funnycompilations8314
9 ай бұрын
that's already ASI level 😉
@jawgboi9210
10 ай бұрын
You mentioned the ability to answer questions about Chinese tones, I'm hoping that Gemini will be able to answer questions about Japanese pitch accent! (similar, but quite different to tones)
@meijiishin5650
10 ай бұрын
That would be nuts. It's way too hard to get reliable information on that.
@deepsp_ce
10 ай бұрын
how about all the different filipino dialects 😮
@raymond_luxury_yacht
10 ай бұрын
Yeah that's hard
@jameslane2326
10 ай бұрын
None of these AI models are worth while to me, until they can remember past conversations, and learn about me, only then will they be at all useful to me. I dont care how smart they are compared to human experts, I only care if they remember the conversation we had last week and if there have been any new updates on the topics we discussed
@maciejbala477
10 ай бұрын
I mean, right now, they are only useful to me on a short-term basis, yeah, which admittedly already helps a lot, but reliable long-term memory is something that would truly make me use AI constantly. Otherwise I just ask it about x or y. I think your point is valid, it's definitely the biggest thing I'd want
@OZtwo
10 ай бұрын
Please when it comes to Google we need to see the finished product here. Remember that Bard had the same technical papers that showed how much better it would be then GPT. Google shot itself in the foot there and no longer can be trusted.
@EmranAmin4
10 ай бұрын
Every time you say you have an announcement I always think you’re about to say ‘Actually guys I was an AI the whole time’
@aiexplained-official
10 ай бұрын
Haha nope
@EmranAmin4
10 ай бұрын
@@aiexplained-official I’m sure your guardrails prompted this response
@aiexplained-official
10 ай бұрын
@@EmranAmin4 Hmm, what if I say, 'I am real, mofo'
@szleventee
10 ай бұрын
Thanks! Keep up the awesome work, it's by far the best channel on AI with a great mix of deep technical insights simply explained! Just great!
@aiexplained-official
10 ай бұрын
Thanks so much! Would love to see you pver on AI Insiders too but regardless very much appreciated!
@watchingvideos9871
10 ай бұрын
Best ai Channel in my opinion. Thank you
@aiexplained-official
10 ай бұрын
thank you!
@Yottenburgen
10 ай бұрын
The MMLU example reeked after having seen you put doubt into it. I didn't even know about them using 32 samples Chain of Thought. If anything I would think Gemini was trained on more MMLU content than GPT4 is too. I hope you're able to benchmark Gemini whenever you get access, your word is far more trustworthy than Googles it seems.
@mahmga1
10 ай бұрын
So grateful you could put this together so fast - Also kudos for calling out the marketing fluff, i.e. the 2 degrees of (im)precision - Embarrassing for the data scientists.
@groovemark
10 ай бұрын
I lov that you highlighted the "cosmic rays" causing delays! Their data center must be too high altitude!
@djJaXx101
10 ай бұрын
I ask Bard if its Gemini and it says "I am not a human and therefore I do not have a zodiac sign. I am a computer program..." After more probing: "My full version name and number is Google AI Bard v1.0.1. I was released on March 21, 2023."
@JohnBr0
10 ай бұрын
9:50 We all knew that cat was never going to make that jump.
@diamondvideos1061
10 ай бұрын
Cosmic rays? Where they using quantum computers 🤣
@jaysonp9426
10 ай бұрын
Thank you! My trust rating for the G company is 1/10 so I'm glad you're calling out the hype.
@lukebenden
10 ай бұрын
Your voice sounds *exactly* like the CEO of Google DeepMind, Demis Hassabis… are you him? 😂
@aiexplained-official
10 ай бұрын
haha no, that would be meta tho
@theaqzer474
10 ай бұрын
Meta would be Zuckerberg @@aiexplained-official
@ShawnFumo
10 ай бұрын
@@aiexplained-official No if it was meta, it'd be a French accent ;)
@randomuser5237
10 ай бұрын
What I find most impressive is how GPT-4V, trained Q2 2022 gets comparable scores for a "natively multimodal model" in Q4 2023 trained on Google's data.
@homemdasneves
9 ай бұрын
you can't forget that ilya sutskever is an actual wizard. now that he left OAI i'm sure that edge they have on everyone else is bound to disappear soon
@yubbityyubba7749
9 ай бұрын
He left? @@homemdasneves
@alphar9539
9 ай бұрын
Gemini is trained in Q1 2023.
@georgesos
10 ай бұрын
If i was the UK government, i d offer you a position in a new AI watchdog and give you all the resources and personnel you might need. No budget limits. You are made for this.
@TheKriss0011
10 ай бұрын
Finally some quality content regarding Gemini! Was refreshing youtube for that!
@almor2445
10 ай бұрын
Did you say the EU is preventing us using and testing this new technology (for our own good) ?
@julius4858
10 ай бұрын
I’ve never once thought about joining someone’s patreon, but here I’m actually considering it. Can you talk some more about the style of content on there? Is it just interviews? Or condensed info like in these videos?
@aiexplained-official
10 ай бұрын
Condensed info. Plus a podcast. Plus personal message and AI Explained bonus videos (will say more in future videos). And lastly will be an AI Insiders Arena where you get to submit an explanation of an idea, code or app that has a chance to feature on the channel. First guest contributors announced soon.
@ethanmuhlestein8187
10 ай бұрын
You are the most thorough and credible AI channel on KZitem for me. I always appreciate your persistent hard work!
@aiexplained-official
10 ай бұрын
thanks ethan, too kind!
@overnightparking
10 ай бұрын
And for this reason I have subscribed to Insiders. Thanks for your hard work and insights!
@cxdxax
10 ай бұрын
Wow, I am more excited about Alpha Code 2, while most attention is on the Gemini Ultra. I have already been using GPT-4 in my daily coding, and I'd say a coder combined with GPT-4 can finish jobs quicker and solve harder problems than working alone. It seems they are exploring a finer collaboration style between coders and LLM. I am really looking forward to it and imagining where this could lead.
@wisdomking8305
10 ай бұрын
But if one person efficiency drastically increase then won't the company only need few highly efficient people. This will result in massive unemployment
@johnnoren7244
10 ай бұрын
@@wisdomking8305 It will and is already happening. World went from a huge lack of software developers to a surplus in just a year or so. This will rapidly spread to other industries and in 1-2 years it should be clearly visible in unemployment rates. Many people don't realize that jobs will be eaten by increased efficiency long before AI fully replace humans.
@davidlovesyeshua
9 ай бұрын
@@johnnoren7244 While I find what you say highly plausible, I am unaware of any present state of a surplus of programmers. And your claims of increased efficiency eating jobs stand in contrast to economic history, where even massively increased efficiency often creates more jobs than it destroys even in the exact industries where the change in marginal output is most pronounced. Coding seems, according to my naïve estimation, somewhere one would expect increased efficiency to result in continued increasing demand since it has so many applications.
@kevinmcfarlane2752
9 ай бұрын
@@johnnoren7244Explains why I’ve been on the bench for such a long time then! 😢 In the meantime, I’m trying to learn AI!
@SeriousGeorge
10 ай бұрын
A lot of people have already stated the same thing, but I also want to congratulate you for creating a great channel out of nothing and earning the credibility to sit down with insiders and experts. All in a very short time frame too! Keep up the good work.
@aiexplained-official
10 ай бұрын
Thanks so much George
@avi7278
10 ай бұрын
I've been watching you since you had like a couple thousand subscribers. I even joked that with such high level content that lacked all the typical click-bait content all the AI grifters use your channel probably wouldn't take off. 200K+ subscribers in like six months later 😆.... you deserve all the success. Cheers!
@aiexplained-official
10 ай бұрын
Thanks so much avi. Unbelievable support, much appreciated and do stick around thru 2024
@MrSchweppes
10 ай бұрын
Amazing analysis! Thanks a lot for the video! I wonder if the release of Gemini Ultra (let's say, it will be in February) will force OpenAI to release GPT-5 earlier than they've planned.
@aiexplained-official
10 ай бұрын
Probably yes, and thanks MrSchweppes!
@MemesnShet
10 ай бұрын
Not long ago i was so excited to run an AI model that contained almost the totality of human knowledge on the internet locally in my laptop now we are getting official support for a state of the art AI model on smartphones,this is amazing I can't wait to try it
@sashank224
10 ай бұрын
Had to rewatch twice, so to summarize, google won this battle for now, until openai fights back. What a impressive LMM. Can't wait to test out alphacode 2 🤩. Since at least for me thats all I can see. Do we have parameter count for Gemini Ultra?
@aiexplained-official
10 ай бұрын
No parameter counts released, though Palm 2 was leaked asap so let's see
@matthewferguson8369
10 ай бұрын
Sorry if I missed it, when will this be available to the public? Gemini Ultra will be beginning of the year?
@aiexplained-official
10 ай бұрын
Gemini Pro available now via Bard (unelss EU or UK)
@reza2kn
10 ай бұрын
OOOOOOF Course this guys learned how to speak Chinese in China! The most impressive thing for me was when they showed how Gemini basically did a systematic review all on its own during "a lunch break", something that takes teams of researchers more than a year to do, collaboratively! So even if you have to hand hold it and account for hallucinations, that's what, day? a week? Why wasn't this around when I was dying while getting my 2 Master's degrees?
@aiexplained-official
10 ай бұрын
Haha epic comment
@AIWRLDOFFICIAL
10 ай бұрын
Can't watch cause im at work but im sure its a great video
@Sk1erDev
10 ай бұрын
But when will my google home actually be useful
@aiexplained-official
10 ай бұрын
Tru
@googleyoutubechannel8554
9 ай бұрын
Just FYI, intrigued by chatGPT 4 coding and CoPilot claims, I put together a 10 question coding benchmark, consisting of questions that are reasonable, (3 friends solved all of them in less than 1 hr), but that are _significantly novel_ compared to what I could find online on 'coding challenge' or on SO sites. (a question is scored as correct if it compiles and returns the answer, and I'm forgiving of environment/version/formatting issues) Here are the code-benchmark results so far. Human 1: 85% (and a long argument about the one she missed :)) Human 2: 100% Human 3: 100% chatGPT 4 Turbo: 0% chatGPT 4: 0% chatGPT 3.5: 0% Copilot codex (GPT3): 0% LLama2: 0% XwinCoder-34B: 0% Bard (10/2023): 0% So let's not get too excited here yet folks.
@kevinmcfarlane2752
9 ай бұрын
A few months ago I ran what I thought were very simple prompts across the main Chat AIs and there were cases where they all failed spectacularly. So we’ve a way to go yet! Nevertheless, I’d still say their impressiveness more than outweighs their weaknesses.
@WeylandLabs
10 ай бұрын
Love the hype of Gemini ! I hope these tech rats get off of Open A.I servers already. Please keep hyping this up... I'll camp Open A.I still. 😂
@jeff01121
10 ай бұрын
Is there going to be a point where the content on this channel will be entirely diverted onto AI insider? Love the content so much that I cannot afford to lose it 🤧🤧🤧
@aiexplained-official
10 ай бұрын
Absolutely not. Much though I'd love for you to join this channel will still be here and strong in 2024 :)
@SaInTDomagos
10 ай бұрын
Now Google can bathe in the warmth of the sunlight for a few weeks until OpenAI bring out their breakthrough model. I have to say, I am very impressed by he specialized models like AlphaCode2.
@RosscoAW
10 ай бұрын
Using the MMLU as a benchmark isn't at all surprising; the issues with it's use are almost exclusively insider knowledge among those who regularly follow AI news, or are involved in the industry sufficiently to understand in-context the relevance of different benchmarks. That is not most people, by a long shot. The reveal of Gemini must cater to and provide sufficient information to all relevant stakeholders (including those with SME, such as AI experts, and yourself [you are without a doubt an expert at AI news aggregation and commentary, hands down]), including normal people who otherwise don't really know or care, and especially mainstream retail and institutional investors (who are more or less just normal, well-intentioned but uninformed laypeople with vastly bigger budgets). The Gemini reveal is above all else a function of Marketing, and Marketing deliverables and contextualization will take priority where possible. I would not categorize their use of the MMLU as a benchmark here as worrying or confusing sign, and simply a function of how difficult it is to accurately and properly market leading-edge AI advancements to a lay public that does not regularly follow the news and developments in the industry more than cursorily or accidentally (by exposure to others in their social circle who might). I would expect that in the future we will receive more relevant and specific information on Gemini, particularly as we get closer to the full release, which is designed more exclusively for the "crowd who follow the AI news and care" rather than the large-scale, public awareness-generating, sentiment building, engagement-hunting nature of a large public reveal to the lay public like this kind of initial press release and tech demonstration is.
@ArSm-ge2qx
10 ай бұрын
Very good comment!
@rihamali6239
9 ай бұрын
it’s my first time watching your videos, and it’s hella interesting and informative, since you have your knowledge in this field what is your perspective on the future tech jobs, cause whenever I go I see people either excited or frightened by this advancement. And for me personally I am excited but a bit worried. Thank you so much again
@aiexplained-official
9 ай бұрын
Hey Riham, thanks so much. I go more into this on my Patreon, AI Insiders, but the short version is I think there will be continued explosion of tech jobs over the next 3-5 years. After that things might get more convoluted, with lower ranked jobs going. Thanks for watching anyway!
@kaikapioka9711
10 ай бұрын
Gemini is far behind )= Kinda disappointed at google but whatever GOOD VIDEO AS ALWAYS :D
@Michael-ul7kv
10 ай бұрын
Exciting look at what's to come. This drills home the need for better benchmarking of AI though.
@bujin5455
10 ай бұрын
15:15. I'm sure AI assisted programming is the future of programming. But for how long? For five months, before the next model comes out with another step change in performance and efficiency? Clearly it's not a future that has any lasting power.
@aiexplained-official
10 ай бұрын
That's my longer term concern. Not for 2020s I feel, partly as the compute power is just not widely available enough.
@alertbri
10 ай бұрын
Superstar, thank you Philip. 🙏👍
@aiexplained-official
10 ай бұрын
thanks alert!
@owen6738
10 ай бұрын
Thank you for the great content. You’re the only KZitemr I have notifications on for
@aiexplained-official
10 ай бұрын
thanks owen, let me know what you think!
@owen6738
10 ай бұрын
@@aiexplained-officialGreat video. The custom UIs that Gemini will be able to create are insane. I’m looking forward to trying that out when it’s available
@blandprix
10 ай бұрын
When he says it's a 30 second ad read but its actually 40 seconds 😭
@aiexplained-official
10 ай бұрын
DAMN!
@iamcoreymcclain
10 ай бұрын
“I read the full 60 page technical report..” THIS is why I wait for your videos. 🤓
@MemesnShet
9 ай бұрын
There is a new interview of Sam Altman on Trevor Noah's podcast where he talks about what happened and he claims nobody left the company hmmmm...not sure if I believe him... Maybe its true as you said the very safety minded employees might have been almost forcibly removed from the company either on purpose or not to give the economic expansion people free roam Tbh even if he's pushing for expansion bigger and better and faster to stay competitive I can't blame him seeing what gemini can do and specially amazing useful integration into existing products It will be hard to compete with that even with the Microsoft partnership
@ce9916
10 ай бұрын
Thank you. Another big AI channel I follow, was just lapping it up. You managed to gracefully point out what needed to be pointed out. Bigger picture, Gemini was barely a few percentage points better than GPT 4 and you know that they cherry picked all the examples. Remind me of the Ernie Bot demo.
@Dart_ilder
10 ай бұрын
Your videos on "let's verify step-by-step" sparked interest in my protein design lab. Thank you for help in staying relevant and informed. I will wait for a follow-up. Also I am very interested on CoT idea "to let the transformers think longer than they are". AlphaFold achieved peak performance partially due to feeding it's embedings back into the transformer block several times. Also this thought of "iterative convergence to the answer" is somewhat similiar to diffusion and flow-matching. May be good idea was reinvented several times. If this isn't interesting enough I would be GREATLY thankful for any redirections for additional info. That's very sad that there are not more AI info youtubers as good as you
@brll5733
10 ай бұрын
Sot it's roughly GPT-4. The real question will be if it's cheaper. That could really put pressure on OpenAI
@Yacobe
10 ай бұрын
The graph from Google you showed is almost Fox News bad in how misleading it is. It looks like the y axis is truncated to 86%. What a joke. That's a pretty big gap for 0.2% difference.
@GrindThisGame
10 ай бұрын
This part from the paper was pretty crazy: Training at unprecedented scale invariably surfaces new and interesting systems failure modes - and in this instance one of the problems that we needed to address was that of “Silent Data Corruption (SDC)” (Dixit et al., 2021; Hochschild et al., 2021; Vishwanathan et al., 2015). Although these are extremely rare, the scale of Gemini means that we can expect SDC events to impact training every week or two. Rapidly detecting and removing faulty hardware required several new techniques that exploit deterministic replay to isolate incorrect computations, combined with proactive SDC scanners on idle machines and hot standbys. Our fully deterministic infrastructure allowed us to quickly identify root causes (including hardware failures) during the development leading up to the Ultra model, and this was a crucial ingredient towards stable training.
@Rick_Mactavish
9 ай бұрын
We've all said that many times but let me repeat myself, you're the best in the field of AI News! Thanks for the video!
@aiexplained-official
9 ай бұрын
Wow thanks Rick
@kylewollman2239
10 ай бұрын
When it's nice outside and I'd rather enjoy the sun than do work, I tell my boss I'm behind due to cosmic rays.
@aiexplained-official
10 ай бұрын
Haha
@groovemark
10 ай бұрын
Why stop at "touch and tactile" training feedback? All sorts of environmental & IoT sensors could be included.
@JohnSmith762A11B
10 ай бұрын
A humanoid robot paired with a fully multimodal model of this quality is basically C-3PO. Can't believe we are this close, this soon. I can easily envision such robots walking around our homes within just a few years. Be honest, did you ever dare think you'd live to see that?
@DavidNorthUK
9 ай бұрын
You're really doing an outstanding job with this channel. This makes all other coverage I've seen on Gemini seem lazy. One observation: I thought the multi modal traning might have some carry-over, giving better common sense reasoning in text prompts but it appears this isn't the case based on the weaker HellaSwag performance. That's kind of disapointing. About the multi-modal capability, do you think its right that gpt-4 isn't truly multi modal compared with Gemini, that they just do image to text in a separate model first? It seems they're claiming to be the first to integrate those modalities fully. In any case its not clear that its giving the benefits they'd like you to believe. I heard that if you give gpt-4 the same prompts used in the video demo (the ones they really used, not what you hear the guy say) then it does just as well.
@infn
10 ай бұрын
Because I'm fully expecting OpenAI to introduce a generational leap soon, this means that Google isn't too far behind. What it does confirm is that there isn't a unique secret sauce that makes ChatGPT unbeatable - there is no moat. It seems to literally be a race to train, do enough RLHF and optimise the models and instances for efficiency. To achieve enough of a competency across all interaction modes and to make it useable at scale.
@ChristianHunter
10 ай бұрын
In which 170 countries is Gemini available?
@mindcache5650
9 ай бұрын
It’s objective but in the now. Just think of the incredible business , health and educational effectiveness and efficiency transformations in 5 years.
@alphahurricane7957
10 ай бұрын
when i saw the community post i thought "well we know what we will watch tomorrow" but you already did a breakdown now. dude you ok? how can you read and watch everything AND do a video in less than half a day? anyway thanks as always :3
@aiexplained-official
10 ай бұрын
thanks alpha!
@MojtabaM-t7q
9 ай бұрын
Hey Philip. I was sad at first that I couldn’t afford your monetized content, because you are the only channel on KZitem that I have notifications on. I watch your content as soon as you post it and I learn and discover a lot from it. The awareness you create is priceless and I hope you make a lot of money for this value you provide. I also hope I won’t miss your valuable information because of payment issues.
@aiexplained-official
9 ай бұрын
Thanks so much my man and having that bell on is enough support already. Maybe one day in a few months or years you can get subbed for a couple of months and binge everything!
@sidnath7336
10 ай бұрын
I do not fully agree that Mathematics will fall first unless if we start training these models specifically by injecting mathematical axioms through symbolic reasoning etc… Mathematics is proof based and not necessarily brute force e.g. you cannot get an AI to brute force the proof of Fermat’s Last Theorem. If it can, then we are truly witnessing the next step in intelligence in the evolutionary chain.
@djJaXx101
10 ай бұрын
I wish the announcement was a bit smoother -New Gemini!!! -its not out yet -but the pro version is kinda in Bard!!! -you can have the ultra later -but not really if your in some countries
@Elonars
10 ай бұрын
Great video as always! If possible, please make a video on custom instructions for GPT-4!
@aiexplained-official
10 ай бұрын
Incredible! Thank you so so much! Have you joined AI Insiders? That's where something more slightly niche like CI would go, even started talking about it in the future of prompting vid, after your comment earlier!
@Elonars
10 ай бұрын
@@aiexplained-official Yes, Philip!
@KillTheWizard
10 ай бұрын
I have never been a big fan of Patreon, but you may be just the person to convince me.
@aiexplained-official
10 ай бұрын
Fingers crossed
@Pabz2030
10 ай бұрын
I dont think you mentioned it but @9.55 is even added humour using a pun to its description of "what happens next" in the cat video. That seems like some serious nuanced understanding of human language
@emrehosgor
9 ай бұрын
Marvelous. What is next embed the model into humanoid and conduct a Turing Test?
@winsomehax
10 ай бұрын
You read all that?. I skimmed the paper, ate some biscuits, tried bard and it hasn't been updated for me (I'm in the UK) and then played with the annoying bing for a bit. I admire your diligence
@newrev9er
10 ай бұрын
Thank you so much for your commitment to explaining these topics! The rate and degree of advances in this domain are overwhelming for a non-expert, like me. I very much depend upon people like yourself to help me understand. As soon as I'm able, I'll be sure to support your work monetarily. I think that, if this technology realizes even a fraction of its speculated potential, then the work that you are doing here will be impossible to overvalue.
@Alexander-dt2eq
9 ай бұрын
soon AI will train humans to perform better :) - we will only rate AI how they can increase performance of humans
@jossefyoucef4977
10 ай бұрын
Couldn't wait for your video on this, thanks for spoiling us!
@procactus9109
10 ай бұрын
That clip at 9:50 is fake.. how does the phone rotate it's screen after the video clip starts but before it's actually rotated while remaining flat on a table ???
@mlock1000
9 ай бұрын
These benchmarks are a bit meaningless. Time will tell what's really going on. A change in the system message will alter the results (more precisely: exactly how you present the query and data) for better or for worse. Shoot, even with exactly the same code for me there are days where gpt4 (api) seems almost useless, and others where it can perform miracles. No idea why. Even crazier, it is literally impossible to prove why this is. One can theorize, and get close to what's happening, but we just don't know what a neural network that has learned things does when it processes data.
@aiexplained-official
9 ай бұрын
Fair point but with every non-text modality pointing in the same direction I think it is very likely to be slightly better
@mlock1000
9 ай бұрын
@@aiexplained-official Don't disagree. It's super cool, with these truly powerful models the only way to know what they are and what they can do is to use them.
@dankodnevic3222
10 ай бұрын
All we can see is Gemini Ultra is better then GPT-4, BUT it is not available yet...
Пікірлер: 653