Igor - I have been following you since the beginning, I promote you to others because you do a fantastic job. Your charisma and enthusiasm are contagious.
Then he sounds like he just drunk a bottle of Vodka 🙂 Like many of his ilk he talks way too fast - shows a lack of ability to understanding the viewers perspective. Its a good video just let down by the speed he rattles everything along at.
@@fx-studiobeing able to speak that quickly so naturally is quite the talent. He’s a pro. I don’t have any trouble understanding him but I’m guessing it takes some getting used to? Hang in there. He’s the best.
🎯 Key Takeaways for quick navigation: 00:00 *🧠 Google's AI Studio is now open to the public, except in Europe, and includes the Gemini 1.5 Pro model with 1 million tokens of context.* 00:27 *🔧 The AI Studio offers more features than ChatGPT, including prompt presets, temperature control, and safety settings.* 01:50 *🧩 The freeform prompt allows you to create prompts with variables and multiple examples, enabling you to build and save custom prompt templates.* 07:48 *🍳 The structured prompt, or few-shot prompting, allows you to provide multiple examples to fine-tune the model for a specific use case, like generating Instagram bios.* 11:29 *📚 The large context window of the Gemini 1.5 Pro model (1 million tokens) enables you to use it for tasks like summarizing long-form content, such as product manuals or podcast transcripts.* Made with HARPA AI
Title: AI can analyze video Proceeds to talk about anything else but no examples of the AI video feature Click Bait Titles 😢generated with ChatGpt 😂 Anyways thanks for the video
Is the data you put in Gemini secure? Would it be safe for a books publisher to use it to summarize their own books without having to worry that the content is going to pop up in other people's chats / responses?
Disappointed. I really thought it was going to parse the video, not just the transcript. People have been pasting transcripts into Claude for a year. Update: It CAN parse a video, but NOT the audio. I tested it with some personal video and it was able to tell me - very generally - what was portrayed in the video. Amazing! But you only mentioned this briefly at 2:38, and never came back to explore it.
Igor; What conference are you attending? Will you attend the AI Bot Summit in Vegas April 22-24? And finally does Gemini allow the creation of your own bots/assistants where you can train it with content and video? Thanks for keeping us updated!
I think Gemini 1.5 Pro's biggest problem is over censoring. The slider to lower the safety settings doesn't have much of an effect based on some brief testing I did. Even during the penguin essay it flagged the prompt as possibly leading to the generation of unsafe content. During casual use of 1.5 pro I noticed it giving very safe non-controversial replies that were kind of useless. This model is made to be safe for children basically. If you prompt it for example with a Psychology text book and ask it to explain narcissism behaviours according to the book, it would probably completely refuse to reply because of a single sentence about suicide being mentioned at some point in an unrelated chapter on depression.
Thanks for sharing your experience, this is really helpful and yeah I am 100 % with you. At the very least the presence of the sliders is a stwp in the right direction
@harmoniics no I didn't specifically do that, but when it's given a large text it will refuse to answer any questions or even read it if it finds anything unsafe or really controversial. For examples I have done, I fed it 3 months of an extensive chatlog with a friend of mine. It was about 700,000+ tokens, and it refused the prompt because about 3 messages were about suicidal topics. I suspected those messages were the cause and changed certain suicidal keywords or deleted sentences, then resent the chatlog, and it answered me then. Another example, I was testing it on the Bible and sent about 35,000 tokens and it refused to give a reply because in one chapter a few people were mentioned to have gotten circumcisions. In that 2nd example I had the safeties turned all the way off
Gemini 1.5's context length is a game-changer for deep data analysis. You can feed it comprehensive market research reports and get granular insights that could refine your targeting strategy.
One of the most helpful and useful videos you've ever made, some people are missing whats really going on here and how powerful organized prompting truly can be for us.
Great video, as usual. Do you know what would round it off? If you followed it up with one explaining, for numties, like me in Europe ( England), how we would use a VPN to get this wondertool! Keep up the good work in transforming an aging troglodyte like me to operate in an AI world.
I've been trying to build an AI chatbot that I can load the 430 pg Davinci Resolve manual into so I can ask it questions about how to do certain things on the fly. This SOUNDS like a good answer for that. However, I uploaded the pdf and after my first 2 questions...both returned an internal error. I've tried this on the paid Claude 3 version, paid GPT, and I think it's just because the PDF is so huge it takes up a mass amount of tokens.
@@deanc5000because we live in North Korea. Do you think that people in North Korea wake up every day and think “hey we don’t know what’s going on in the world”? No. They have been oppressed and censored for so long that they just don’t know anything different. It’s the same for most people in Canada. We live in a dictatorship behind an increasingly corrupt and authoritarian regime. We live behind an information Iron Curtain. If people don’t make the effort to learn about the rest of the world they just think it’s “normal”. That’s how communism works. Did you vote for any of the 100’s of legal codes, acts and statutes that got rammed through parliament in the past two years? Did you vote for Bills C 11 and 18? It’s literal censorship. The media doesn’t educate us. They try to brainwash and indoctrinate us into anti American tax slaves. The propaganda machine is in full swing. I think most people realize something is wrong but have not connected the dots. The algorithms on yt have changed a lot recently. They really want us to watch CBC 😂😂. Ain’t gonna happen! Canadian Brainwashing Corporation Gawd I wish canukistanians would wake up… but it’s too late now. We are going down big time. I doubt if we can vote our way out now. Most born Canadians are too stoned on SOMA to care. New Canadians are grateful to be tax slaves… likely worse where they were born? Sorry… ranting 😔 This channel is one of the best on TH-cam and I should not be ranting. Sorry. I love AI and Igor is truly exceptional …. I hope that google and anthropic can find a way to help us. Elon is trying. I think lots of good people are. Unfortunately most good Americans are also dealing with a lot of evil forces trying to bring them down.
Hi, thanks for your video, anyway sorry but this is not the first to analyses TH-cam videos, the Gemini extensions already do this, also has another utility that does this and it does first than Gemini extensions, and has another platform that does this, so this also does not uses the audio. Anyway I liked your video and thank you very much since I learned with your videos, but this is not the first, just another way to analyses videos
Why are people saying you talk too fast? Smh...it's weird how some people just like to complain. There are Speed functions on the videos if you cant hear well or keep up. Pretty simple
If it is biased but based on our own data, then doesn't it just show through the AI that Modern Standards have also become biased. You Know what I mean, like if it were trained on only Soviet Russian Era Stuff it would be however it is because of its training data. So maybe it's a sign to people to stand back look at all these supposed equality acts and ask if they are truly equal to all peoples.
Have you figured out what is going on yet? No anthropic. No Gemini now? We are the beta testers for all of the un WEF davos elites. But who really knows for sure?? I know nothing… but things have changed a lot recently. About every 80 years something happens. 🤔. I’m not bored! 💞🦋. Do we live in a simulation? Wes Roth seems to think so. Maybe it doesn’t matter?
If nothing else i hope other AIs use this "censorship slider" so that users can control it. Maybe even you know a simple "i am over 18" checkbox to turn off any of that kind of censhorship. Doesnt change the fact that the political censorship will always be there but oh well.
Why you never talk about open source AI tools? I guess it will be probably because you want to take profit of what you do with this channel and you want big corps like google invite to give lectures and get paid for it but, have you even think that if the control of AI rely on big corps is the worst case scenario for the future of humanity?
Hmm interesting that you say that because I do talk about them in the friday videos its just that the closed ones are currently the most powerful, so that‘s what I personally use and then teach.
Love the video Igor … not sure how you stay on top of all this . Wish I had a v p nnnn. Or? “I AM a man” 😂👏 💯 Nice to see that the LLM overlords walked back on the weird algorithm and is giving people options. It’s ok… everyone is moving to open source. Fascinating. Great to see you IRL at the conference. 🙏☀️
Bro you're going too fast, this coupled with your accent make it difficult for anyone to follow you, especially for a nonnative English speaker, like me. .
These are great practical examples along with even better Insights for why they work. I need to spend more time with Gemini.
Agreed, BUT no sooner do you start to gain understanding than a new feature pops up making the old almost obsolete.
I do too! Feel like I've barely tapped the potential of this tool
As a 10+ years Mac user and full stack developer, the biggest lesson I learned in this video is [Command+Shift+V]😂
😂 That‘s a win
Igor - I have been following you since the beginning, I promote you to others because you do a fantastic job. Your charisma and enthusiasm are contagious.
for those of you saying hes talking too fast, go to settings and slow down the speech rate. Also keep closed captions on. Just want to be helpful
Then he sounds like he just drunk a bottle of Vodka 🙂
Like many of his ilk he talks way too fast - shows a lack of ability to understanding the viewers perspective.
Its a good video just let down by the speed he rattles everything along at.
@@fx-studiobeing able to speak that quickly so naturally is quite the talent. He’s a pro. I don’t have any trouble understanding him but I’m guessing it takes some getting used to? Hang in there. He’s the best.
@@CM-zl2jwyes
I uploaded a video and it told me what was inside!
Then, I asked it can it analyze video and it said no LOL
🎯 Key Takeaways for quick navigation:
00:00 *🧠 Google's AI Studio is now open to the public, except in Europe, and includes the Gemini 1.5 Pro model with 1 million tokens of context.*
00:27 *🔧 The AI Studio offers more features than ChatGPT, including prompt presets, temperature control, and safety settings.*
01:50 *🧩 The freeform prompt allows you to create prompts with variables and multiple examples, enabling you to build and save custom prompt templates.*
07:48 *🍳 The structured prompt, or few-shot prompting, allows you to provide multiple examples to fine-tune the model for a specific use case, like generating Instagram bios.*
11:29 *📚 The large context window of the Gemini 1.5 Pro model (1 million tokens) enables you to use it for tasks like summarizing long-form content, such as product manuals or podcast transcripts.*
Made with HARPA AI
stinky europeans?
Title: AI can analyze video
Proceeds to talk about anything else but no examples of the AI video feature
Click Bait Titles 😢generated with ChatGpt 😂
Anyways thanks for the video
Thanks for the heads up. Won't continue watching it then :D
Is the data you put in Gemini secure? Would it be safe for a books publisher to use it to summarize their own books without having to worry that the content is going to pop up in other people's chats / responses?
thanks for this -- would love to see a dedicated video on how to use as a student/teacher 🙏
Thank you for this video! This is really helpful!
Respect to video editing team for illustrating "Stop sequence" 😂😂😂😂😂
Disappointed. I really thought it was going to parse the video, not just the transcript. People have been pasting transcripts into Claude for a year.
Update: It CAN parse a video, but NOT the audio. I tested it with some personal video and it was able to tell me - very generally - what was portrayed in the video. Amazing! But you only mentioned this briefly at 2:38, and never came back to explore it.
Igor; What conference are you attending? Will you attend the AI Bot Summit in Vegas April 22-24? And finally does Gemini allow the creation of your own bots/assistants where you can train it with content and video? Thanks for keeping us updated!
I think Gemini 1.5 Pro's biggest problem is over censoring. The slider to lower the safety settings doesn't have much of an effect based on some brief testing I did. Even during the penguin essay it flagged the prompt as possibly leading to the generation of unsafe content. During casual use of 1.5 pro I noticed it giving very safe non-controversial replies that were kind of useless. This model is made to be safe for children basically. If you prompt it for example with a Psychology text book and ask it to explain narcissism behaviours according to the book, it would probably completely refuse to reply because of a single sentence about suicide being mentioned at some point in an unrelated chapter on depression.
Thanks for sharing your experience, this is really helpful and yeah I am 100 % with you. At the very least the presence of the sliders is a stwp in the right direction
ok.. but did you prompt it with a psychology book and ask it explain narcissism according to the book? I was just curious
@harmoniics no I didn't specifically do that, but when it's given a large text it will refuse to answer any questions or even read it if it finds anything unsafe or really controversial. For examples I have done, I fed it 3 months of an extensive chatlog with a friend of mine. It was about 700,000+ tokens, and it refused the prompt because about 3 messages were about suicidal topics. I suspected those messages were the cause and changed certain suicidal keywords or deleted sentences, then resent the chatlog, and it answered me then. Another example, I was testing it on the Bible and sent about 35,000 tokens and it refused to give a reply because in one chapter a few people were mentioned to have gotten circumcisions. In that 2nd example I had the safeties turned all the way off
Awesome- thank you so much!
There is an AI tool called "AI Writa". Could you please give us your review about it? Like what's the pros and cons of this AI tool.
Gemini 1.5's context length is a game-changer for deep data analysis. You can feed it comprehensive market research reports and get granular insights that could refine your targeting strategy.
One of the most helpful and useful videos you've ever made, some people are missing whats really going on here and how powerful organized prompting truly can be for us.
Thanks , long context window queries are expensive with Opus and Open AI Assistants. Going to lake a look at Gemini Pro!
Great video, as usual. Do you know what would round it off? If you followed it up with one explaining, for numties, like me in Europe ( England), how we would use a VPN to get this wondertool!
Keep up the good work in transforming an aging troglodyte like me to operate in an AI world.
Pick an east coast US node and let us know! Or is it blocking major VPNs...?
Awesom as always!
Great video 👍
I wonder if Gemini 1.5 Pro overloads and crashes if there aren't minorities in the video.
auto kicks you from the platform and permabans your google account 😂
Im perma-banning myself from Google.
OpenAI need to wake up ... amazing
any idea when google ai studio will be available in uk/europe.....or shall i just pay for a vpn?
I can't see how to access the freeform prompt?
Great content
Great Information!!
very good
Hi Igor love form India I am following you since many months loved your work
same here
yo that clip scared the shit outta me man lol. Today and YESTERDAY OMg
This is able to replace a subscription for many people.
If I am not a developer, can I still use it? And how much it cost?
So, the features are only relevant for Pro users who pay for pro versions?
I've been trying to build an AI chatbot that I can load the 430 pg Davinci Resolve manual into so I can ask it questions about how to do certain things on the fly. This SOUNDS like a good answer for that. However, I uploaded the pdf and after my first 2 questions...both returned an internal error.
I've tried this on the paid Claude 3 version, paid GPT, and I think it's just because the PDF is so huge it takes up a mass amount of tokens.
Have you tried adding your document to a GPT? Last time I checked, you could upload 20 PDFs.
Not in Canada :(
Google hates us up here.
Canadian government is smoking sum
Anthropic not in Canada either. Would like to know why??
@@deanc5000because we live in North Korea. Do you think that people in North Korea wake up every day and think “hey we don’t know what’s going on in the world”? No. They have been oppressed and censored for so long that they just don’t know anything different. It’s the same for most people in Canada. We live in a dictatorship behind an increasingly corrupt and authoritarian regime. We live behind an information Iron Curtain. If people don’t make the effort to learn about the rest of the world they just think it’s “normal”. That’s how communism works. Did you vote for any of the 100’s of legal codes, acts and statutes that got rammed through parliament in the past two years? Did you vote for Bills C 11 and 18? It’s literal censorship. The media doesn’t educate us. They try to brainwash and indoctrinate us into anti American tax slaves. The propaganda machine is in full swing. I think most people realize something is wrong but have not connected the dots. The algorithms on yt have changed a lot recently. They really want us to watch CBC 😂😂. Ain’t gonna happen! Canadian Brainwashing Corporation
Gawd I wish canukistanians would wake up… but it’s too late now. We are going down big time. I doubt if we can vote our way out now. Most born Canadians are too stoned on SOMA to care. New Canadians are grateful to be tax slaves… likely worse where they were born? Sorry… ranting 😔 This channel is one of the best on TH-cam and I should not be ranting. Sorry. I love AI and Igor is truly exceptional …. I hope that google and anthropic can find a way to help us. Elon is trying. I think lots of good people are. Unfortunately most good Americans are also dealing with a lot of evil forces trying to bring them down.
i think copilot can analyze videos (youtube videos) when i play the video on edge and ask copilot , question about the video.
Is there a free api for this too?
Another amazing TH-cam video Igor 🎉🎉 thank you ❤❤
I don’t know how to use it instructions unclear and don’t see the image and video option this is so clickbait
Hi, thanks for your video, anyway sorry but this is not the first to analyses TH-cam videos, the Gemini extensions already do this, also has another utility that does this and it does first than Gemini extensions, and has another platform that does this, so this also does not uses the audio. Anyway I liked your video and thank you very much since I learned with your videos, but this is not the first, just another way to analyses videos
Why are people saying you talk too fast? Smh...it's weird how some people just like to complain. There are Speed functions on the videos if you cant hear well or keep up. Pretty simple
Maybe it‘s the different microphone (travelling for a conference). Arguably my home setup has higher quality audio. Appreciate the comment though :)
@@aiadvantage yes sir. Top quality stuff. I think this may be my new LLM option. Thanks
If it is biased but based on our own data, then doesn't it just show through the AI that Modern Standards have also become biased. You Know what I mean, like if it were trained on only Soviet Russian Era Stuff it would be however it is because of its training data. So maybe it's a sign to people to stand back look at all these supposed equality acts and ask if they are truly equal to all peoples.
You didn't talk about Europe
Huh, they omitted Canada, yet 106 other countries have access.
Have you figured out what is going on yet? No anthropic. No Gemini now? We are the beta testers for all of the un WEF davos elites. But who really knows for sure?? I know nothing… but things have changed a lot recently. About every 80 years something happens. 🤔. I’m not bored! 💞🦋. Do we live in a simulation? Wes Roth seems to think so. Maybe it doesn’t matter?
Did you try a VPN?
here in Asia 😎
If nothing else i hope other AIs use this "censorship slider" so that users can control it. Maybe even you know a simple "i am over 18" checkbox to turn off any of that kind of censhorship.
Doesnt change the fact that the political censorship will always be there but oh well.
It‘s better than nothing but I agree with you. I covered completely open models like Dolphin which give you the completely uncensored LLM.
ChatGPT can analyze video through GPTs
have you used any that worked particularly well? reply with some names if so, would love to cover those too
Why you never talk about open source AI tools? I guess it will be probably because you want to take profit of what you do with this channel and you want big corps like google invite to give lectures and get paid for it but, have you even think that if the control of AI rely on big corps is the worst case scenario for the future of humanity?
Hmm interesting that you say that because I do talk about them in the friday videos its just that the closed ones are currently the most powerful, so that‘s what I personally use and then teach.
It's this free?
yeah! pretty amazing tbh
Love the video Igor … not sure how you stay on top of all this . Wish I had a v p nnnn. Or?
“I AM a man” 😂👏 💯 Nice to see that the LLM overlords walked back on the weird algorithm and is giving people options. It’s ok… everyone is moving to open source. Fascinating.
Great to see you IRL at the conference. 🙏☀️
Forth comment 🎉
Nobody cares. At all.
B B h chi
this feels like some kind of secret code...I'ma see what Gemini makes of it
What ever you do DON'T get tattooed.This awful fad is leaving the West a visual mess !
Bro you're going too fast, this coupled with your accent make it difficult for anyone to follow you, especially for a nonnative English speaker, like me. .
noted, thanks for the feedback friend 👍
Can you just talk a bit slower? i am over 30 years old.