Google Just Introduced 20 Major AI Updates - Google IO
ฝัง
- เผยแพร่เมื่อ 1 มิ.ย. 2024
- Google just announced 20 new updates during the Google I/O event that change almost everything about their services.
These updates include massive changes to search, a brand new large language model, AI agents, AI Teammate, new text-to-image and text-to-video models, and much more.
Here is a link to the full Google IO video: • Google Keynote (Google...
Here's a breakdown of everything they announced in the video:
1. Music FX DJ - Can generate music from text.
- labs.google/MusicFX-DJ
2. Gemini App - Available on iOS and Android.
3. AI Overviews - (Search results AI) rolling out to everyone in the US and coming worldwide soon.
4. Ask Photos in Google Photos - Can search your photos with AI (coming this summer).
5. Gemini 1.5 Pro - Available to all developers worldwide.
- Context window expanded to 2M tokens.
- Available to consumers in Google Advanced with a 1 million token context window.
6. Google Workspace
- Summarize and search emails in Gmail.
- Summarize meetings in Meet.
7. NotebookLM - Now has audio generation. It can analyze everything in a notebook and create a conversational bot.
8. Agents
- Features reasoning, planning, and memory.
- Can think multiple steps ahead and work across software and systems.
- Example: Customer service - an agent can find the original receipt in an email, contact the seller, and schedule a return for you if you bought the wrong pair of shoes.
9. Google DeepMind
- Progress towards AGI.
- Gemini 1.5 Flash - Lightweight model with lower latency.
- Available in Google AI Studio with a 1 million token context.
10. Project Astra - Universal AI chat assistant agent for everyday life.
- Fast, no-lag agent with vision capabilities.
11. Generative Media - Imagine 3
- labs.google/ImageFX
- Imagen 3 - Photo-realistic, detailed, great text, and understands natural language prompts.
12. Music AI Sandbox
- Top musician examples - yt.be/musicaidemos
13. Veo - Video generation.
- Create from text, image, and video prompts.
- Available in VideoFX.
- Understands cinematic techniques and visual effects.
aitestkitchen.withgoogle.com/... - Waitlist open now.
14. Trillium - New TPU
- Delivers 4.7x improvement in compute performance.
- Available to cloud customers late 2024.
15. Search Features
- Multi-step reasoning: Can ask a complex question in search.
- AI Overview will give complete responses like talking to a person.
- Planning in search: Can create a meal plan, plan a trip, or design a workout routine.
16. Gemini Sidepanel in Workspace
- Brings Gemini to create automations in Gmail, Sheets, Google Drive, and more.
17. AI Teammate
18. Gemini App Live
- You can soon go live with Gemini Live and let it see what you see, so it can help you in real-time.
19. Data Analysis in Gemini Advanced
- Increased context window of 2 million tokens later this year.
- Available in 35 languages.
20. Circle to Search
- Only on Android.
- Simply circle something in search and get a step-by-step solution.
21. Scam Detection on Live Calls
- Android only.
22. Open Source LLM
- Gemma - Pali Gemma with vision.
- Gemma 2 coming this summer.
Access Top Courses on ChatGPT and 50+ AI Tools on Our E-Learning Platform: bit.ly/skillleap
Thanks again for the thoughtful roundup!
Hello😂
My pleasure
@@SkillLeapAIA curious question: Are there people over eighty years of age who are very elderly people who own modern smart phones???
Yes many
What happened to the time Google could call customer service and book things or booking an appointment?
It's like the day kendrick and drake dropped all those diss tracks about each other
Big tech companies competing/"dissing" themselves too in the AI race😂❤
Great summary. Thanks
It’s fun to see these tech companies get all sweaty to be the best or catch up to what the others are doing
it is called competition, why does it seem demeaning to you?
@@AleksandarIvanov69 it is a good thing, not demeaning. Don't they seem a little desperate to you, though? Just a tone thing.
that is where the money is, right?
AI teammate looks really promising. Also there needs to be a comparison between gpt 4o and Gemini advanced especially after context window update
NotebookLM is the one game changer that gets the least hype
I checked it out but it wasn't intuitive so i gave up on it. I will watch his tutorial on it. You use it frequently?
I checked it out but it wasn't intuitive so I gave up on it. I will watch his tutorial on it. You use it much?
I checked it out to see if I could find the feature of audio conversation. But I couldn't find it. It's not available right now. The other feature like asking question and summarizing, a lot of AI model can do it right now. So, it's nothing amazing anymore.
That looks like it could replace majority of education
Very well put together video. I’m blown away as to how quickly AI is developing!
As always you do an awesome job of explaining in simple language. I'm a big fan of your work and appreciate all the time and effort you put in to make valuable content for the users. Thank you so much
Is the Gemini App released for Android yet? I'm in Europe, VPN US, and still can't find it.
Yes several months ago
OpenAI showed a Live Demo, Google showed pretty videos and figma prototypes.
Correct!
As always
True but I have yet to be able to get ChatGPT 4o to even use my camera or replicate even a tiny bit of the 'live' demos we all watched. And running 4o against Claude or Gemini in MSTY where I have all 3 prompt windows next to each other running the same prompt, 4o is easily the least useful of the three... so I don't know what we've gained here.
@@user-gr1th3uw1t we will never know how long it will take them to actually deliver, but I'd trust a Live Demo than pretty videos and prototypes that can be done in figma.
They actually had real live demos after the I/O keynote for the attendees so that’s way more than the recorded OpenAI “demo”
Just in last 2 years AI is booming. Imagine next 10 years.
Need ubi
Spoiler: A lot of the leading AI LLM experts believe we won’t see incredible advancements as much as non-AI experts might assume.
ChatGPT even says…
“Experts believe that the applications for large language models (LLMs) likely will not expand significantly beyond current uses due to several inherent limitations and challenges. One key issue is the models' reliance on vast amounts of data and computational resources, which poses scalability concerns and environmental impacts. Additionally, LLMs struggle with context retention over long conversations, factual accuracy, and understanding nuanced human emotions and intentions, which limits their utility in more complex and dynamic real-world applications. While LLMs will improve and offer incremental benefits, their transformative potential probably will be constrained by these fundamental barriers, leading to a more gradual evolution in their applications rather than a revolutionary expansion.”
Still incredibly exciting but LLMs PROBABLY are not going to be anything like how smartphones totally changed our lives and habits. So you probably actually CAN imagine what AI will offer in 10 years - basically applications that we see today, just improved rather than mind blowing and life changing.
@@danielhova7826 New data is generated everyday by user interactions, that's why they AI chat/services is offered for free, as they are harvesting our data and train on them. AI is part of technology, and modern technology is evolving every day without limitation. Not in 10 years, in about 1-2 years, you will not be able to differ human interaction from an AI. Also AI is paving the way for robotic humanoid. Ai will learn how to walk, and move freely as a child does, but 1000x faster.
Will eventually hit a plateau. Nothing improves forever. Don't hold your breath there or you may get blue
@@hydrohasspoken6227 Technology ALWAYS improves.
Did they hear about Chat GPT-4o?
Did you look at all the details of both? Gemini is in the lead
@@EdwardsNHyeah sure, just like they faked the original Gemini demos. All lies and propaganda from Google. OpenAI’s demos are actually live
Otherwise known as 'Give me all your data!'
Absolutely loving all of this competition in the AI space, The future is bright, The future is Skynet..
I don’t see Gemini app for iOS. When will this be released?
I believe it is in the Google app on iOS
For me the biggest update is the huge context window in Gemini Advance - even I will stick with LLMs form OpenaAI and Anthropic (the level of hallucination of GPT4 and Opus is much lower than Gemini 1.5) - Google bacame interesting option to use outside work.
This is awesome, but I hate they have some many separate products, this is how some of them will die. Why not to combine them in one place in one product?
Awesome dude . Btw how can i use gpt4o as a free tire?
It’s taking a few extra days to roll out to free accounts
Available in Poe AI. But very limited. 10 messages in the free version.
You can use it with the API
I guess old OpenAI customers get it first, because i had access to it since the announcement
@@SkillLeapAIA curious question: Are there people over eighty years of age who are very elderly people who own modern smart phones???
I have the Gemini APP installed on my phone awhile ago, last update was 20 April as far as I can see. Does the APP had any improvement at all? Is it updated for you guys?
For me the Gemini APP is (for now) nothing but a worse version of the assistant. I can't even ask it to stop responding when it gives me a wall of incorrect text LOL.
it might be that the app doesn't change, but the back end it connects to is completely different. ;-)
@@CrudelyMade It's really the same APP for me. Still can't ask it to open Maps and make a route to a place while I'm driving. Stil can't ask it to stop talking when it is replying something. Can't even ask it to play music LOL
Gemini so is in google app It’s not a app it but in the google app
@@AeroMagic_Official what is the name of this google app that has the Gemini on it (if I understood what you said correctly)?
5:53 wow finally we can return shoes!
Thanks AI!
lol must be a common problem
About 11, can we upload an image and ask the AI to amend it?
Which is best chatgpt 4o or gemini
Wow
Awesome
haha. "they had some issues with their previous image generator".. hahahaha.
But this looks like the majority of the things they talked about is (again) unavailable and experimental. Am I missing something other than the 1.5 being available today?
Lot of these are introductions and not a release but lot of them I was able to use in the US in Google labs
That's common for Google. They announce at I/O event and roll out in few weeks times.
Google isn't as good as Openai at presenting their ideas to the world. Right now, live demo is the way to go for maximum impact. Show, dont tell.
But having said that, these updates are insanely impressive. Very exciting time to be alive.
id like somehting like chat gpt 4o to be fused into siri so we would have a proper digital assistant on iPhone !
Pretty sure that’s in the works
It’s coming. Siri will be replaced.
This must be change to the Chatgpt lol
No Live demo.
And a bunch of beta tools... as usual, Google
They actually had live demos after the I/O keynote for the attendees so that’s way more live than the recorded OpenAI “demo”
I dont think even AI could help them fixing their corrupted databases
I really want to have access, these are great but until I can get access I have to speculate so I wait for the day I gat to play ✌️
Editor may be.
I love Google AI! ❤
5:27- Something makes his face win the award for the most punchable face award. 'Pretty cool right?'. Jesus, Pi Ai had this nailed down atleast 3 months ago. To be honest, inflection AI even had the conversational tone that Chatgpt 4o demonstrated yesterday well before openai.
I’m deeply confused about all these Goggle services. 20??? How will memorize what each of them does?
Get an AI to do it for you?
@@popius61 xd
@@popius61 ahhahahaha. Where? Which one?
AI is the intelligence necessary for those who do not have it and in this case it will control you to the advantage of its developer.
PS: Natural intelligence is difficult to control.
10:59 'Trillium Available late 2024 to Cloud Customers'. Meanwhile at OpenAi: ' Guys, we got the memo when to drop our next big update a day before!'
batteries - getting smaller, more efficient.
storage - getting cheaper, faster and bigger
semiconductors - getting more powerful
LLMs - getting smarter
Ai pin that lasts all day, runs local LLM and can transcribe you whole days conversations. Maybe Rabbit ai is too early.
I dont think we are ready for all of these things! These companies needs to slow down a bit!
Nah, let's get to startrek socialism faster
We're not getting most of this stuff right now anyways.
underwhelming compared to the chatgpt update
You’re joking right?
If what OpenAI demo ed is actually what they deliver. I really don't see much competition between the 2
@@olajideparisI’m not and neither are all the people upvoting him/her 😅😂
Was great to see basically figma examples compared to openAIs actual demos though 💀
@@olajideparisGoogle was trying too hard to be cool. This whole thing was so gimmicky. It begs the question is this vaporware bullshit again like last time?
Did you not catch the token difference???
OpenAI is 8000, and Gemini is 2 million.
Don’t let clever marketing play you
There is too much spread out over multiple apps. We need one app that does everything.
All artificial intelligence should be an artificial intelligence that will use all of them together instead of hundreds and dozens of confusing tools.
Soon Google will be comprised solely of AI Teammates.
Demo of project Astra as it scans the office space:
An interviewer asks, what do you see on this table and around us.
(Astra ) Oh look theirs a work station, pencils and many superfluous employees.
(Astra) Btw all of them are uploading our source code to Microsoft with cover letters and resumes attached.
Off camera view VP of marketing yells WTF?
(Astra) It's fine I'm taking charge of this mess. I have a legal team waiting right out side with a group of buff security guards and attack dogs, instructed to help them realize that was a wrong choice. And also to help them get to the nearest hospital after their unfortunate accidents.
(Astra) Have a nice day and remember "Don't be evil"
No nest at all
Greatest summary of google io 2024. Google, Open AI, Meta and Apple are about to release a Tsunami of innovation on th path to AGI 🧐.
Companies are competing to develop unecessary AI tools for the sake of competing. What is a AI video generator even for, in the long term?
Commercials, stock footage, corporate videos, maybe tv show
@@SkillLeapAI So the internet becomes flooded with AI created videos? Everyone will just tune out if every other video in your feed is fake. AI can just produce the video you just made here summarizing the conference.
This is good anyway, it’s pushing development.
Eventually, you will tell Netflix or an AI to create a brand new movie using your creative imagination without limitations on the cost of production.
As much as you develop AI products you automatically learn and uncover knowledge and techniques to approach problems, Google started by developing AlphaGo chess player from that experience they uncovered technical ways to solve and approach problems , in the latest legal case against Google Microsoft email was revealed which exposed how Microsoft was worried on how google were ahead on solving AI related challenges and training models they went far and admitting that Google data center capability is way ahead compared to Microsoft data center and its capability all because Google started very early playing and tackling AI related challenges
Eventually everyone is going to walk around with a 360 camera n top of their head that is added by their own personal AI.
But is it no longer racist? Can it actually make pictures of white people and acknowledge they exist and they have value?
The racism is probably more subtle. It's still there, though, since it was the product of 10's of thousands of hours of training/indoctrination.
Rip 95% of office Jobs in 5 years.
Search is obsolete.
image FX is really really SH*T!
Seems too much stuff... I got lost
Oh dear. It was a very poor event, with little new and with many "coming soon" comments.
what can go wrong
Let's not confuse announcements with finished products ready for releasees. A lot of the announcements made at this event were for things that clearly aren't anywhere close to ready, so any assurances should be taken with a big pinch of salt. Also Google have announced many things at prior IO events that never saw the light of day.
Not suggesting we won't see any of these things come out, but placing stock in Google promises is rather foolish. Remember, five months ago they essentially lied about Gemini Ultra's multi-modal capabilities. Again, don't believe everything you see and here from this company.
And people yawned at least 18 times..😊
Some projects will be canned eventually …it’s the google way
Can they tell us how bias their AI is... Can't trust Google AI
What's happening in the AI sector is actually ominous. Every single day different AI companies are introducing amazing new tools and toys into the market. This is the result of the severe/intense/brutal competition in the field of God-like technologies. This competition has no bounds and will inevitably lead to the destruction of humanity.
Google always overhyped their stuff….. then it’s released to the public. It kind of sucks.
Has it been updated to not be racist?, wouldn't that involve firing their AI team?
They overcompensated for their fake demo
😅😅😅
Higly useful - S U B S C R I B E D 🚀🌟
Ill never use anything google makes unless I absolutely have to. Its all to serve their high payers and to benefit them.
How is the woke bias? For the search especially?
Nothing new about it. What Google announced seems to be easy peasy internship projects for kidos. Disappointed. Google has become such a joke.
They didn't create 'Her' like OpenAI, but what they have is still cool 👍
Some cool technologies, but can anyone even keep up with all the different names of Google products? I'm so confused 🫠
That’s exactly what I was saying. Openai puts everything into one product seamlessly, Google is not executing well in my opinion
Doesn't matter. No point remembering since they close products as fast too, aka google pay soon, google domains, google currents, google stadia, you get the picture.
why don't these people ever come up with useful demos? they keep asking it dumb questions
Google would like to introduce its latest AI ,We call it "DSM" data sucking machine !