If Gemini does a good job at acting as a personal secretary with knowledge from Gmail, Calendar, Docs, Drive, Keep, and Maps. I'll actually consider paying for the service.
It's a new modality. Keeping an eye on this. I'll wait a few weeks after Pixel 9 is released, to see if there are any problems, before considering getting it. (AI is always a bit finicky. But I've gotten enough value out of it, and have learned to manage my expectations.) There's a little bit of latency. (But this probably can't be helped for any serious workload.). Seems like a ChatGPT-4o voice competitor. Though OpenAI has yet to have this widely deployed. I can certainly understand this, given the compute that is needed to support this, esp. in a wide deployment. It seems like Gemini Live will be amongst to first to widely release this.
This one is different. It'll be part of the OS than just a cloud based chatbot. Thing is only Apple can compete with Google with the iphone. OpenAI doesn't have the hardware.
@@h.c4898 Apple likely doesn't have the backend. 7 Data centers vs 37. Microsoft does but Microsoft doesn't have mobile phone hardware. They might have to join together to be able to compete with Google.
@@DavidKnowles0Microsoft failed on the mobile space. There is one winner and loser and the rest. MS is in there. Only Apple can compete with its Apple intelligence.
Nice but... It still has the unnatural behaviour that many or all AIs have, the parrot síndrome as I call it, I mean if I ask you "What time is it?" your answer will be...ten thirty and not "So you want to know what time is it, atm it's ten thirty am WEST (Western European Standard Time)" or something like this
Slowly robbing people of their humanity. As if the loneliness epidemic wasn’t bad enough-soon people will be saying their phones are their best friends.
If that Gemini live feature runs on servers which I presume it does, then it would be catastrophic for the planet, energy consumption wise. I totally reject Google’s carelessness in attempting to normalise such a destructive feature into every day consumer life.
Let's hope they teached it math cos from my experience (with the free versions) chatGPT is way "smarter" in maths and it gets what we ask him with way less prompts than Gemini
I was one of the lucky members of "the public" to get access to Gemini Live in a non-controlled setting. My initial experience: The good: It can answer questions that can be answered by directly typing them into google and looking at the first result. The voice sounds natural enough, but with a hint of robot, precisely as demonstrated in this video. Latency is pretty good, as good on the "public" side as seen here. Interruptions can be handled, its just not perfect. It can handle sarcastic input, and doesn't go off the rails if the user isn't "fully nice" (not shown above). I have not tested it on actual nonsense extreme rudeness or profanity, because that's not a legit use case. The bad: The model powering the voice is far weaker than normal Gemini, can only answer basic questions, fumbles on anything that would require a "person of slightly higher than normal intelligence to answer". Not all of your voice inputs are accurately fed in, reading the transcript it often misses stuff (which explains some of the nonsense) When you run into "content issues" (for benign stuff that false positive trips the censor), it'll just stop answering and appear to glitch out.
Cool but the problem is you're cutting constantly when showing the AI... which yeah isn't gona work for me. It needs to be 100% zero cuts to prove or at least help prove that it wasn't edited.
Someday we'll find out that Google hired 10000 people to answer Gemini questions same as Amazon's AI-based 'just walk out' checkout which was powered by 1,000 Indian workers manually reviewing cameras 😂😂😂
The Google event was honestly a total flop. The presenters were all nervous, kept stumbling over their words, and then one of the Pixels straight up didn’t work, so they had to swap it out live. The whole thing just felt way too stiff. And don’t even get me started on the products-they’re all just screaming “AI” and desperate for attention, like they’re trying so hard to stay “innovative.” I’m so bored of all this. Seriously, could you guys shock us with something actually innovative for once? AI is super dull and usually totally unnecessary.
I'm currently using this. It's nice but not as good as Gpt, which is still to be released. G. Live sounds robotic. And with just a male voice. How can to make this more dynamic and with a female voice.
It’s so chatty. It’s like a RPG video game dialogue you just want to skip but are afraid to because you might miss the one nugget of info that’s important to the story. Exhausting.
3:18 The fluid response between each interruption is very naturally sounding. And the accurate response to the subject in discussion each and every time is smooth, this also make it sound truly natural. The ability to corespondent accurately is flawless and fascinating.
All true, but if I remember correctly, OpenAI is a lot faster and more fluid and much better at emotion. We will see who reaches wide distribution first. Supposedly, OpenAI is in limited alpha last I heard.
@@gavinderulo12 end of fall, really? Interesting. As I understand it open AI has been delayed because of security stuff: looks like now Google will be the one rushing ahead with something insecure, but maybe not, as googles new voice mode doesn’t seem nearly as powerful as the open AI advanced voice mode.
What a waste of time with all those greetings and back and forth. From a computer I was a list of options that I can evaluate in a glance and pick the most trustworthy one. I don’t want the computer’s opinion.
PS: So much money spent in advanced options while basic ones are left behind like if it's going to replace G.A why doesn't it already do at least the same as it, is it that hard to copy, is it that hard to at leas make them talk with each other, I mean if it's taking this long for one to replace the other at least make them talk to each other or something meanwhile, I wouldn't mind if I could at least say "Hey Gemini tell/ask G.A to turn my kitchen lights off"
Despite my speech stutter, Gemini Live is somewhat able to understand me; I hope they implement a feature in the future to better understand people with a speech stutter.
I am using it in the app, and a replacement for Google's voice assistant on my S24 Ultra. This is a total gamechanger in respects to quality of life.. this helps me so much as a chef, photographer, and marketer, that is constantly juggling work/devices.
without prejudice, a AI system might actually be good at allowing people to practice their social and conversational skills, especially when they get to the stage of being able to mimic different personalities.
It doesn’t seem to be multimodal like GPT 4o but she didn’t ask it to sing or talk really fast or slow or make animal sounds so it’s difficult to know for sure. I wasn’t really impressed.
@@FriedChairs it's not multimodal, but it is an AI voice assistant. I don't need theatrics, I don't care if it sings, I need to be able to voice commands for recipe conversions, setting timers, responding to texts, the weather, any question that pops in my mind. Also where is the GPT4o voice model? I was a subscriber to OpenAI since Day 1 but their bogus marketing hype put me off.
Sounds and behaves pretty meh. Nothing we haven't seen before. Competitors are way ahead in their game. Pi, a free alternate is miles ahead of this thing.
Pretty cool. But it takes too long. I don't need/want to have a conversation. Just need bullet points. If you could do it that way, I think that would be something I'd use. Also, when it comes to something like product reviews and such, I think it would do VERY well. Takes reviews by top people gives good and bad and overall thoughts. You can fine tune it for someone who doesn't understand the technology, or ramp it up to someone who is a tech nerd. That way based on person's level of understanding it'd be able to "talk" to you in the best way for you to understand.
From what I'm seeing, that's just the initial latency, when first connecting to the live feature. After that first delay, it should be faster.. Plus giving demos at events like this, the wifi is already slow.
@@TrentonMatthews yes for sure, but we have voice mode like this on PI, character ai, .... It's just the regular voice mode on chatGPT with voice activity detection and streaming STT
I saw an interview of Rick and he said they're planning to put Gemini everywhere. Like your phone experience will be Gemini. What about what the customers want? I don't want a crap load of AI features and assistant every damn where
I love having this voice assistant as a replacement for Google Voice (which I used extensively before LLMs). I have it setup on my S24 Ultra and I can just fire questions at it from the lockscreen.
