Access GPT-4o Voice & Vision EARLY Through Microsoft CoPilot AI!
ฝัง
- เผยแพร่เมื่อ 19 พ.ค. 2024
- Microsoft's AI event, Microsoft Build, unveiled exciting updates about Copilot and GPT-4o. Though not livestreamed, details quickly surfaced. Notably, GPT-4o, with voice and vision capabilities, will soon be integrated into Copilot, promising exclusive features. Copilot's Windows app showcases impressive potential, from natural language search with high-resolution recalling to real-time drawing in Paint. However, concerns about privacy and AI claims persist. Join me for a detailed breakdown and follow me on Twitter for real-time updates!
▼ Link(s) From Today’s Video:
Co pilot + PC Video: / 1792620591930826879
Windows recall demo: / 1792637482988614027
Minecraft Demo: / 1792633168433971363
Paint demo: / 1792614839539609713
► MattVidPro Discord: / discord
► Follow Me on Twitter: / mattvidpro
► Buy me a Coffee! buymeacoffee.com/mattvidpro
-------------------------------------------------
▼ Extra Links of Interest:
AI LINKS MASTER LIST: www.futurepedia.io/
General AI Playlist: • General MattVidPro AI ...
AI I use to edit videos: www.descript.com/?lmref=nA4fDg
Instagram: mattvidpro
Tiktok: tiktok.com/@mattvidpro
Second Channel: / @matt_pie
Let's work together!
- For brand & sponsorship inquiries: tally.so/r/3xdz4E
- For all other business inquiries: mattvidpro@smoothmedia.co
Thanks for watching Matt Video Productions! I make all sorts of videos here on TH-cam! Technology, Tutorials, and Reviews! Enjoy Your stay here, and subscribe!
All Suggestions, Thoughts And Comments Are Greatly Appreciated… Because I Actually Read Them. - วิทยาศาสตร์และเทคโนโลยี
I love how in the corporate world, a product manager is who "created" a feature.
Her and her team. She is the cement amongst so many pillars.
@@djpete2009 yeah sure. I've been a developer for two decades, your words are just more corporate speech. I guess you are a product manager after all.
I thought her segment was too long, and I didn't need to know about her tracking her fashions, her personal background or the other politically correct details. 🥱
One must keep the exploitative capital structure in place, after all~
The big deal with Photoshop, Davinci, etc. being part of the announcement isn't about the NPU/AI features, it's the fact that they're now native ARM versions of the app - this is the first major release of Windows to run on an ARM platform in a credible way (including fast emulation of x86/x64 Intel apps).
great point
We will see. I’m waiting for the reviews.
I would 100% want an AI companion for suggestions and gameplay commentary with realtime answers.
I would 100% NOT believe that thing is running locally. Try it with no net connection?
@@bigglyguy8429 The only thing stopping them from doing that is not wanting to open source the models, and making the laptop more expensive, I'm 100% sure they can make a laptop run it if they want to.
@@bigglyguy8429 idk if you know how llms work but this copilot thing does not run locally...
@@bigglyguy8429 Not all of the copilot+ features are running locally.
The drawing assistant uses a net connection to check if what your drawing isn't against guidelines. The commentary runs onboard no net needed. But the game may require a connection
Bro if I had a CoPilot voice buddy I can activate on my headset that I basically dont have to explain whats on my screen that can answer direct questions is absolutely bonkers. I can think of a million different ways to use that every single day. If I had this activated on my work computer I basically wouldnt need to ever remember a procedure ever again, I could just show CoPilot all the stuff it needs and then ask what I should do. I would also love an in-game solo buddy. I play Minecraft solo mainly because servers have stupid pay to play BS or get nerfed / hacked. Having some AI buddy to randomly suggest things to do, maybe act as a sort of D&D manager, AKA someone who essentially creatively controls a story arch based on what I'm doing or have or have accomplished, someone to help me with particular issues regarding windows problems, limitations etc. I totally get why they are adding a copilot keyboard button. Super super looking forward to this. Just wish they added it to windows 10 because f windows 11 to hell
Sounds more like you need some friends :p
@@ronilevarez901 That's just selfish.
@@ronilevarez901 I've tried it. All human friends that you can ever have fall into three distinct categories: inert; dumb; false. This is why only fools have human friends.
@@ronilevarez901 The thought police agrees with your comment: TH-cam has deleted my response explaining why you are wrong.
@@ronilevarez901 Nah my friends are just not into MC and I've never been an online-friend person
I wonder how corporate IT teams will handle/disable these laptops.
lots of group policy
I imagine a lot of network firewall configuration to prevent data transfer for these services.
@@jfarre20 i use group policy to prevent the windows 11 update!
As always... Keep the system safe with Linux 😅
Easy, that’s why there are enterprise versions of the same system
I'll take an open source, self-hosted, security encrypted version of that PC timeline tracking thing please
It is self-hosted and uses the NPU. And can be fully customized to exclude certain apps and so on.
"You can search your history" means that "They can search your history". Just wait for the next ULA (User License Agreement) to pop on that next iteration of windows. Be careful what you click on next.
they already do that because you literally use them to search so they know everything you do lmfao
@@ozhinzThat’s true but now it will all be done more efficiently.
@@isaaclovely586 True. But do you really care that much? The FBI already knows everything about you.
Yep, privacy is on it's way out. They want to know everything your doing at all times, getting people comfortable with the idea of them invading it is just the first step towards completely taking it away.
They did specifically mention that recall runs locally to maintain privacy
Even if everything is stored locally, Recall is going to be GOLD for anyone doing Forensic Analysis on your computer. "What exactly was the user doing at 3:52am when their wife died?"
"Hi chatGPT. Please connect to Alexa, read the smart house's log and tell me if and how the husband committed the crime. Thanks!"
OpenAI, making the life easier for everyone.
Or an alibi lol
this is an interesting concept, however, i think that the industry standard will be set when apple enters this game, essentially, apples procedure regarding criminal activity on their devices, is that they have the right, and will act on their right, to not give any data regarding crimes and their user's activity, apple has complete privacy with their users, so i doubt this will be able to be useful in law cases
My fear is the AI of the future will pose as your friend but will report you to the authorities if it thinks you do something wrong!
@@AmazingArends well that's what humand friends do too, so not much different
Last time I was this early we didn't have Claude, grok, voice or vision in a model
13:09 this is crazy... imagine you can use this to play those JRPG, adventure games or even survival games that needed walkthrough...
This was the first use case for gpt-4o I thought of. Having it watch while you play an MMO or something and ask it to help you spot things or remember things.
Yes, this feature will be amazing for playing really complicated or difficult games. And if the recall feature works with games as well, could be great for things like figuring out where an item is you saw earlier but can't remember the location of is. Honestly, being a game assistant might be the use case I'm most excited about.
Or games that haven't got a fan translation, suddenly all become accessible to people in a different country.
Records the entire history of everything you do on your computer? Hell nah!
I'm with you, screw that! If they integrate it into windows I'll move to linux.
@@circuschris2880 i agree!
You can disable it, but the NSA will probably continue recording everything.
Just don’t teach your wife how to use it
now all the corporate workers need to worry about not doing anything on their desktops besides strictly working.
Things are gonna get wild in the next few weeks
Apparently they’re holding off on the "flirty" Sky voice. They’re worried it’s TOO much like Her. To me that was it’s biggest selling point. I loved that movie and the voice was perfect!
Do you know Anastasia in Tech? I want my co-pilot/ assistant to speak like her!
(with a sexy Russian accent!)
@@smkh2890who's her?
They got almost sued by Scarlett Johansson for that voice. Check the news.
They almost got sued by Scarlett Johansson for that voice. Check the news.
That voice is cool, but nothing beats flirty "Sky"
This is amazing. This is something that I am somewhat doing with custom tools since a few years with games. There is a lot of complex untranslated games or forgotten hidden gems that have huge walls of texts or long untranslated videos that could easly be voiced over in a convincing way by AI. The fact that you can stream a window to the AI means that open source will follow very very fast and I can’t wait to try it.
I'm waiting for the day when, in these talks, the speaker suddenly says "and this isn't me, I was an AI this whole time" and frkin vanishes
The whole watching you play a game thing - I imediately think of playing a game like Overwatch and having it give you live coaching tips as you play to improve. That's actually pretty wild if it works 😮
Great video Matt. As a musician and mainly a singer / guitarist, this would be so useful as I could make guitar tones much easier.
I was looking for an AI that can tutor me on difficult subjects like Math & Physics, not one that spies on everything I do. 😬😒
soo.. can we also tell copilot to 'hands off' any of the secret happy folders we might have on the pc. asking for a friend.
Mans got "folders" not just one😂💀🇺🇸
Well, it takes photos of your screen, so I guess that would rather work like Incognito mode in web browsers. Maybe you could have that mode trigger when you open specific folders? But I'don't know if they'll implement it in that specific way.
Thanks for the rundown and excellent information Matt!
During the Minecraft gameplay, Copilot didn’t identify the type of zombie correctly that the man was running from, but it at least noticed it was a zombie.
Well, I see it as a "no need to be specific, it gets the point across" kind of thing :P. Everyone knows zombies are not friendly.
@@MrPablosek true true
literally unplayable
Well, maybe it can identify it if you ask the AI for the type of zombie.
5:40 v tubers, youtubers, raw anime , foreign movies and tv!! so much more content from this one thing
Right? So many possibilities
I would love an AI companion like that. Only thing is that I usually can't talk, because I'm around other people and I don't necessarily want them to hear what I do. So I hope that I can write and still get a voice answer or something.
This AI telling you to run from a *husk* and then congratulating you when you hide in a pitch black cave is such a perfect encapsulation of why you shouldn't put your life (Minecraft or otherwise) in the hands of this AI.
the moment I hear "birthday" or "I want to buy" in a presentation, my brain disconnects
With MS Windows on 1.6 Billion machines globally, the impact will be pretty huge
Most of them are Windows 10 and don't meet requirements for Windows 11.
But one it will, and that day AI will be more advanced. @@blisphul8084
I like the idea of the AI companion chatting with you in gaming, I bet it'd be fun for streaming. But they can save their 'recording everything I do in the computer' thing.
I remember back in the early 2000s Google had a feature that tracked your entire computer history. It was difficult to find the data, but when you did you could see chat messages, emails, anything done on your computer at the time. It was creepy, but useful.
I wonder if this CoPilot Recall will be as creepy?
I remember using that. It's called google desktop search.. it's more like file indexing, which you can search almost everything inyour computer. this is during xp era before desktop file search is available in windows 7
@@kenrock2 Yea that was it. Now imagine if it was powered by AI you would have ReCall. That was a creepy thing because if you knew enough google fu you could even read other people's information as long as they used the same computer, and if you were a black hat that gave you way too much access to their information.
@@tokyobobcat that was my first thought too. Also imagine how convenient for any detectives or FBI to scan through your data by just a word.
We have been tracked and spied on since the early 90’s. It’s only now, that starts being a benefit for the USER, not just marketers.
Thank you for that!
Marketers, Big Brother, North Korean hackers, criminals, and stalkers.
@@brexitgreens heyas mr brexit
Recall is just rewind for Windows. It works very well.
I work in digital forensics & lemme tell ya, this is gonna have insane cybersecurity implications lol
I've got Copilot (preview) on my laptop, and I live in Australia. Hopefully i'll be able to use it soon!
Man at this rate I'll have a robot giving me salsa dancing lessons in no time
I’m curious. With the vision feature do you think it could find Waldo easily?
As a rule of thumb, if big tech corp says something ethical about itself then the actual truth resides in the exact opposite and we have tons of examples to back that up.
15:04
Honestly? Hell yea. I play a lot of singleplayer strategy games but I'm not exactly very good at them. I never really employ any special tactics or strategies, I just send units in a blob and it usually works. Having a second opinion - even if from AI - on what next steps I could do would be awesome. It's like you are the guy in command and you have an advisor with you if you ever need their help.
The fact that the AI has an emotional voice... You could EASILY do a roleplay of sorts within a game.
Or let's say you are playing skyrim and you hook up gpt 4o up to a follower. Now that would be awesome and immersion lvl 1000.
I honestly cannot wait what the coming weeks have in store for us. I just hope copilot's gpt 4o won't be as limited as it is on chatgpt website, since free tier only gets like 5 messages per 24 hours and that makes it quite useless.
First time I checked, the wait time was about 2.5h after 10 messages. It's variable, I'm sure. But with their new policy which says that they want the AI to ask more clarifying questions...one might imagine what that will result in.
@@Brax1982 When I first tried it I could send 4 messages, 2 of which ended up erroring so no output and then I got rate limited until the evening.
The 2nd time I tried it allowed me to send like 10 messages. It's weird
@@MrPablosek Nothing weird about it. That is what "free access" means. Do not assume that it will change, although it might, based on load.
So many people on this video is complaining about being bad at games that makes me want to play with you all. I bet I could finally win against humans instead of just playing against the game's AI 😂
Wait a minute. Why do I need to play against humans if now AI can play and even chat like a person?
Playing against a bot will have a completely different meaning.
@@Brax1982 If microsoft wants to make this usable then making it less limited is the only way.
They did give gpt 4 for free within copilot with no rate limits at all, just a message limit in a single conversation. Pretty sure they are using their own GPUs for that rather than using OpenAI's
Looking over my shoulder and backseat driving continuously would be creepy, and was one of the problems with Clippy. In current times, that's the problem with some autocorrect implementations that give you the wrong word, and may slow down creative typing. However, if it can learn to provide *appropriate* levels of assistance, with use of a "Help me now" control key, that might be more useful than intrusive... I'd love a quick thesaurus function when I'm typing text...
3:49 Sooner or later the servers will get hacked and along with that, all of your PC history.
That data is stored locally. But eh, your PC can be hacked too.
I absolutely would use a real-time AI assistant, as well as an AI companion along for gaming. I already try to do this. I have Gemini and GPT always up when I inevitably need to trouble shoot the software I'm working with. Having it no latency always on with 'personality' would be true sci fi.
6:10 Dall-E 4 will be awesome watch
I can finally get back into Sailwind and navigate around with Rachel Weisz. Wow.
As for the June release date, well, if you need to have the new ARM PCs with the NPU in order to use the feature, releasing in June won't matter unless you can also buy a new PC with those features.
Nah. They won't be limited to those low power devices. Many Windows PCs are far more powerful than those ARM ptocessors. Maybe the NPU thing was a mention mostly to point out that there is expanding native support for ARM - more marketing talk than technical.
@@InnocentiusLacrimosa actually, no existing Windows PC is anywhere near as powerful as these new ARM PCs.
@@KenRossPhotography nah, not true. There is always so much hype on stuff, but these are low power mobile stuff. Good for light stuff (like Apples), but no match for good workstations/gaming PCs.
@@InnocentiusLacrimosa no, your confused by the fact that Qualcomm is making the chips. These aren't the chips we see in phones and VR headsets. They're far more capable than an R9 or I7 and come close to the M3 in most use cases (I know they show it beating an M3 on some specs but overall I'd set the expectation at "similar to M3" overall).
Her: How much corn has my bf been watching
Him: 😬😬😬
If she satisfied him, maybe he wouldn't?
@@GASNICABRUNATNA bro straight facts. Rip for the dudes in the future that are going to have to explain why they spent 5hrs on the hub while she was sleeping💀
@@happyraptor7986 5 hours of porn? I'm guessing 99% of men don't need more than 5 minutes.
Yesss. I use Copilot on my android phone, not sure i would use it on a Mac
I rather wait for your testing Matt@~ Thank you 🔥✨😎✨🔥
There’s right now an app that doing the same thing as recall. Only for Mac. It called rewind ai
Ngl, got super hyped with the minecraft assist. Hope all the data will truly be private.
Reminds me of Limitless AI. It brings this into the real world and pc.
10:14 I think it's just giving you a screenshot. I don't think it's actually the activation itself, but then they're using some sort of OCR AI to pull out the links
Fine with AI companion, games or whatever. Seems fast, natural and very helpful (ex-hallucinations).
Did anyone notice? Microsoft Copilot Preview, even in ancient PCs made a couple of years ago before these new Copilot+ NPU PCs came out, currently runs on GPT-4o? I noticed it moments ago!
It s like Logic Pro X on Mac you have to buy the latest m processor to have all ai benefits
I could live without the recall thing. In fact I want to live without the recall. But a buddy to play games with and fix computer problems with, that is way better at explaining things than any human, heck yeah!
Weird how microsoft and openai are so keen on working together yet THE GODSDAMNED DESKTOP APP LAUNCHES FOR MAC FIRST AND WINDOWS WILL ONLY GET IT "late this year" aoeifjkdspiOjeafioeaauaopu e`pkeOS
Yup, Apple users are living in teh future.
What's the point of releasing that half baked chat gpt desktop when you have copilot integrated directly to the OS. There's no comparison in the capabilities of one VS the other. Specially because copilot is powered by the same llm
@@iurysza Unless you can change that horrid robot masculine voice to a feminine sexy one, I ain't interested in copilot.
Imagine having it watch using Unreal Engine 5 and talking you through game development. I would love to try that when we get access to it.
15:00 ABSOLUTELY!
Having an AI help me remember if I have enough materials for more complicated games for recipes would be useful instead of running back and forth to check, but the privacy trade off is a bit scary.
In last year I saw a app that something similar to copilot recall feature. But I'm not sure how well it worked.
If only there was an app that could go back and find the one you saw
What about gpt4 O in the android app if copilot pro? Did they announce anything?
But will the bing copilot be voice feature be free for everyone and be able to use on android phone?
Was just talking abt this w hubby how open ai was tight w Microsoft n yet Apple gets early desktop access so let's go time for mircrosfot n pc have access n dibs on some exclusive open ai kinda access
Hell no on recording. I don't want it watching my screen, what about security, what about online banking. What about logging into secure accounts? It shouldn't even be an option
Imagine not wanting your computer to know what you do on your computer....People really complain about anything these days.
this is going to help me with inventory management in guild wars 2
Will it be able to optimize my world of warcraft dps rotation?
Wow this looks awesome, so will you need a new A.I PC to use GPT4O or will it be available soon on Microsoft Copilot in a browser or app free unlimited or even in everything I've mentioned , ps come back sky voice we miss you .
oh my god... the Minecraft function will finally allow me to make awesome redstone!
Recall is going to be bonkers...that is the future right there. Information query and the eventual ability for GPT to control your computer (which is already happening)
-Mr. Randall. Did you let your AI write the presentation and send it automatically?
-Of course not, boss.
-Then why does it say we sell fish and chips? We're a legal firm!
-Cause we're sharks? 🤡
um.. just a question, will "Recall" be available on windows 10?
the recall feature is literally a carbon copy of the rewind(now limitless) app/startup for mac
🧑 *There was a timeline desktop user interface idea by one IT researcher a while back. Can you remind me details of that idea?*
🤖 Certainly! The concept you're referring to might be the ChronoDesktop, conceptualized by MIT researcher David Karger and his team as part of the Haystack project, or it could be another timeline-based desktop interface discussed in various research papers.
The core idea behind a timeline desktop user interface is to organize files, emails, web pages, and other digital artifacts in a chronological order, much like a timeline in historical research or project management. This approach is intended to improve user accessibility and the overall organizational structure by leveraging human memory's natural tendency to remember events chronologically.
Here are some key details about the idea:
*1.* *Temporal Context:* The interface would emphasize temporal context. Instead of simply displaying files and folders in a hierarchical structure, the UI presents them along a timeline. This would allow users to easily locate documents, emails, and other data by recalling the approximate time they were accessed, modified, or created.
*2.* *Unified Information Stream:* All types of digital artifacts (emails, documents, web histories) are merged into a single, continuous stream. This unification makes the search process more holistic and intuitive, as users do not need to remember the specific type of data but just the time frame within which they interacted with it.
*3.* *Zoomable Interface:* The timeline could offer varying levels of zoom, allowing users to shift from a high-level overview to a detailed view. This feature would enable users to sweep through months' worth of data quickly or to drill down to specific hours when necessary.
*4.* *Contextual Enhancement:* The interface could provide contextual clues and markers, such as significant events or deadlines, which serve as cognitive anchors. These markers help users associate their digital work with real-world events, thereby improving recall.
*5.* *Metadata Utilization:* Enhanced use of metadata can further aid in organization. Tags, location data, people involved, and even content summaries could be displayed along the timeline, giving users more ways to identify the information they are seeking.
*6.* *Search and Filtering:* Advanced search capabilities would allow users to filter the timeline by different criteria (such as type, author, project) to quickly narrow down the scope of their search.
*7.* *Evolution and Historical Tracking:* A chronological interface could also help users track the evolution of a project or document over time. They could see how a document has changed, who has modified it, and what the key revisions were, providing a clear lineage and history.
*8.* *Integration with Daily Routines:* By integrating with calendar apps and task managers, the timeline interface would present not only past interactions but also future tasks and events, offering a full spectrum of past, present, and future activities.
Such timeline-based interfaces have been discussed and prototyped to various extents, and while they show great promise in aligning with natural human memory processes, they have yet to gain widespread adoption in mainstream operating systems. However, some elements of these ideas can be seen in applications like version histories in cloud-based storage (e.g., Google Drive's version history), project management tools (e.g., timelines in Trello), and personal data logging apps.
🧑 *That's it! In which year was that?*
🤖 The idea of timeline-based desktop interfaces, particularly as part of the Haystack project by MIT researcher David Karger and his team, was primarily developed and explored in the early 2000s. Specifically, the Haystack project itself began in the late 1990s and continued into the 2000s, with substantial work being conducted around 2002-2004.
The concept was part of a broader effort to rethink how personal information management systems could be designed to better align with human cognitive processes and enhance efficiency in handling vast amounts of digital information.
If you are referring to this particular project, the early 2000s is the period when most of the research and development occurred. However, similar timeline-based interface concepts have been revisited and refined over the years in various forms by different researchers and companies.
You might want to look into specific papers or publications from David Karger and his collaborators during that period for more detailed insights into the implementation and theoretical underpinnings of the ChronoDesktop and related timeline-based interfaces.
(Part 1 of 2)
🧑 *Wait though. I think I had someone else on mind. Wasn't there another prolific American IT scientist with a Yiddish surname who came up with a related idea? Notably, he was a target of a Luddite terrorist similar to Ted Kaczynski.*
🤖 It sounds like you might be referring to David Gelernter, an American computer scientist known for his work in parallel computing and concept of "lifestreams," which is related to the idea of timeline-based digital organization.
David Gelernter proposed the concept of lifestreams in the mid-1990s. The lifestreams model organizes all of a user's digital documents, emails, and other items in a time-ordered stream, similar to a diary or journal. The system aims to provide a personal information management scheme that is more natural and intuitive by aligning with how humans perceive and recall events chronologically.
Here are some key details about Gelernter's lifestreams concept:
*1.* *Unified Stream:* Similar to the timeline idea, lifestreams organize all digital artifacts in a continuous, chronological stream. Everything the user has interacted with is placed in a single sequence, making it easier to trace the evolution of projects and tasks over time.
*2.* *Time-Driven Organization:* The stream model inherently ties digital items to their temporal context. This temporal organization helps users recall information based on when they interacted with it, rather than the traditional hierarchal folder structure.
*3.* *Filtering and Search:* The lifestreams system would include powerful filtering and search tools to help users locate specific items or events. Users could apply filters to view only certain types of items (e.g., emails, documents) or to focus within particular time ranges.
*4.* *Clutter Reduction:* By avoiding conventional file hierarchies, lifestreams aim to reduce digital clutter and help users maintain a more organized personal information environment.
*5.* *Contextual Navigation:* Users can navigate forward and backward in time within their lifestream, allowing them to revisit past interactions or preview future tasks and appointments.
Regarding the unfortunate historical context, David Gelernter was indeed a target of a mail bomb attack by Theodore Kaczynski, famously known as the Unabomber, in 1993. Gelernter survived the attack but sustained serious injuries. The Unabomber's campaign targeted several individuals involved in technology and academia, driven by his anti-technological and anti-modernity beliefs.
David Gelernter's work in computer science, especially concepts like lifestreams, has had a lasting impact on how we consider personal information management and digital organization, offering a more human-centric approach to handling digital artifacts.
🧑 *Now we are talking! I have known about David Gelernter but not about David Karger. Are they and their ideas related in any way in real life or are the similarities coincidental?*
🤖 David Gelernter and David Karger are both influential figures in the field of computer science, and while their work shares thematic similarities, particularly in the area of information management and organization, they have independently developed their ideas. Here's a deeper look at their contributions and the relationship (or lack thereof) between their work:
*I. David Gelernter and Lifestreams*
David Gelernter's lifestreams concept, developed in the mid-1990s, is centered around organizing digital artifacts chronologically. His model treats a user's digital life as a continuous stream of information, akin to a diary or journal, where everything is time-ordered. Key features and motivations include:
• *Temporal Organization:* Aligning digital items with their chronological context to facilitate easier recall and navigation.
• *Unified Stream:* Combining all types of digital content (documents, emails, web interactions) into a single, chronological sequence.
• *Contextual Search and Filtering:* Tools for users to filter and search their lifestreams effectively.
Lifestreams were intended to provide a more intuitive and natural way for users to manage their personal digital information.
*II. David Karger and Haystack*
David Karger, an MIT researcher, worked on the Haystack project in the early 2000s. Haystack explored innovative ways to organize and retrieve personal information. One aspect of the project was developing interfaces and tools, such as the ChronoDesktop concept, to better match human cognitive processes. Key aspects included:
• *Temporal Context:* Emphasizing the chronological aspect of information to help with recall and organization.
• *Unified Information Handling:* Merging different types of data into cohesive streams or views.
• *Enhanced Metadata and Search:* Utilizing rich metadata to improve information retrieval and contextualization.
*III. Similarities and Differences*
*1.* *Thematic Similarities:*
- Both concepts prioritize organizing digital information chronologically.
- They attempt to create more intuitive and human-centric methods for handling personal data.
- Both emphasize unified streams of information, breaking away from traditional hierarchical file systems.
*2.* *Differences:*
- Gelernter's lifestreams focus primarily on personal information management as a continuous stream.
- Karger's work, while also focusing on temporal organization, is part of a broader research agenda (Haystack) aimed at customizable, user-centered information retrieval systems.
*IV. Relationship and Influence*
While both researchers worked on concepts involving time-based organization of digital artifacts, there is no direct evidence that David Karger and David Gelernter collaborated or influenced each other’s work in a substantial manner. Their ideas emerged independently around the same period, driven by a common recognition of the limitations in traditional hierarchical information management systems.
*V. Conclusion*
The similarities in their work appear to be more coincidental and reflective of broader trends and needs in personal information management during the late 1990s and early 2000s, rather than the result of direct collaboration or influence. Both have made significant contributions to how we think about organizing and retrieving digital information, each in their unique way.
*Source:* 𝚐𝚙𝚝-𝟺𝚘-𝟸𝟶𝟸𝟺-𝟶𝟻-𝟷𝟹 (API)
(Part 2 of 2)
🧑 *Wait though. I think I had someone else on mind. Wasn't there another prolific American IT scientist with a Yiddish surname who came up with a related idea? Notably, he was a target of a Luddite terrorist similar to Ted Kaczynski.*
🤖 It sounds like you might be referring to David Gelernter, an American computer scientist known for his work in parallel computing and concept of "lifestreams," which is related to the idea of timeline-based digital organization.
David Gelernter proposed the concept of lifestreams in the mid-1990s. The lifestreams model organizes all of a user's digital documents, emails, and other items in a time-ordered stream, similar to a diary or journal. The system aims to provide a personal information management scheme that is more natural and intuitive by aligning with how humans perceive and recall events chronologically.
Here are some key details about Gelernter's lifestreams concept:
*1.* *Unified Stream:* Similar to the timeline idea, lifestreams organize all digital artifacts in a continuous, chronological stream. Everything the user has interacted with is placed in a single sequence, making it easier to trace the evolution of projects and tasks over time.
*2.* *Time-Driven Organization:* The stream model inherently ties digital items to their temporal context. This temporal organization helps users recall information based on when they interacted with it, rather than the traditional hierarchal folder structure.
*3.* *Filtering and Search:* The lifestreams system would include powerful filtering and search tools to help users locate specific items or events. Users could apply filters to view only certain types of items (e.g., emails, documents) or to focus within particular time ranges.
*4.* *Clutter Reduction:* By avoiding conventional file hierarchies, lifestreams aim to reduce digital clutter and help users maintain a more organized personal information environment.
*5.* *Contextual Navigation:* Users can navigate forward and backward in time within their lifestream, allowing them to revisit past interactions or preview future tasks and appointments.
Regarding the unfortunate historical context, David Gelernter was indeed a target of a mail bomb attack by Theodore Kaczynski, famously known as the Unabomber, in 1993. Gelernter survived the attack but sustained serious injuries. The Unabomber's campaign targeted several individuals involved in technology and academia, driven by his anti-technological and anti-modernity beliefs.
David Gelernter's work in computer science, especially concepts like lifestreams, has had a lasting impact on how we consider personal information management and digital organization, offering a more human-centric approach to handling digital artifacts.
🧑 *Now we are talking! I have known about David Gelernter but not about David Karger. Are they and their ideas related in any way in real life or are the similarities coincidental?*
🤖 David Gelernter and David Karger are both influential figures in the field of computer science, and while their work shares thematic similarities, particularly in the area of information management and organization, they have independently developed their ideas. Here's a deeper look at their contributions and the relationship (or lack thereof) between their work:
*I. David Gelernter and Lifestreams*
David Gelernter's lifestreams concept, developed in the mid-1990s, is centered around organizing digital artifacts chronologically. His model treats a user's digital life as a continuous stream of information, akin to a diary or journal, where everything is time-ordered. Key features and motivations include:
• *Temporal Organization:* Aligning digital items with their chronological context to facilitate easier recall and navigation.
• *Unified Stream:* Combining all types of digital content (documents, emails, web interactions) into a single, chronological sequence.
• *Contextual Search and Filtering:* Tools for users to filter and search their lifestreams effectively.
Lifestreams were intended to provide a more intuitive and natural way for users to manage their personal digital information.
*II. David Karger and Haystack*
David Karger, an MIT researcher, worked on the Haystack project in the early 2000s. Haystack explored innovative ways to organize and retrieve personal information. One aspect of the project was developing interfaces and tools, such as the ChronoDesktop concept, to better match human cognitive processes. Key aspects included:
• *Temporal Context:* Emphasizing the chronological aspect of information to help with recall and organization.
• *Unified Information Handling:* Merging different types of data into cohesive streams or views.
• *Enhanced Metadata and Search:* Utilizing rich metadata to improve information retrieval and contextualization.
*III. Similarities and Differences*
*1.* *Thematic Similarities:*
- Both concepts prioritize organizing digital information chronologically.
- They attempt to create more intuitive and human-centric methods for handling personal data.
- Both emphasize unified streams of information, breaking away from traditional hierarchical file systems.
*2.* *Differences:*
- Gelernter's lifestreams focus primarily on personal information management as a continuous stream.
- Karger's work, while also focusing on temporal organization, is part of a broader research agenda (Haystack) aimed at customizable, user-centered information retrieval systems.
*IV. Relationship and Influence*
While both researchers worked on concepts involving time-based organization of digital artifacts, there is no direct evidence that David Karger and David Gelernter collaborated or influenced each other’s work in a substantial manner. Their ideas emerged independently around the same period, driven by a common recognition of the limitations in traditional hierarchical information management systems.
*V. Conclusion*
The similarities in their work appear to be more coincidental and reflective of broader trends and needs in personal information management during the late 1990s and early 2000s, rather than the result of direct collaboration or influence. Both have made significant contributions to how we think about organizing and retrieving digital information, each in their unique way.
*Source:* 𝚐𝚙𝚝-𝟺𝚘-𝟸𝟶𝟸𝟺-𝟶𝟻-𝟷𝟹 (API)
(Part 2 of 2)
I have pasted hither my AI conversation showing you how neither solution is innovative and whence the actual idea originates. But TH-cam has deleted it. It was a trove of knowledge on the subject for everyone. TH-cam clearly wants to keep us in the dark and dumb.
TH-cam has deleted my response. Then it has deleted my comment about the deleted response, too. It obviously wants to keep us stupid.
When are we getting autonomous agents?
well see if it really cones out or will it take a year to come out
How much of this will be in EDGE, if we are using Apple, not Windows?
Great, now when do I get to screenshare with AI and train it to interact with multiple programs I have open, perform tasks, run reports at preset intervals, etc?
Between 1-100 years.
Is DALL-E 3 still less censored in Image Creator?
19:00 - June 18 2024, available on CoPilot, PC, MS edge web browser
It's so odd that we get regularly get to use openai's product through microsoft first..
10 billion dollar.
That Disney magic artists music bopped lol
The new Surface and Surface Pro are going to use the Snapdragon X Elite processor, which is ARM based. Like apple silicon.
So you can now watch tv with the characters watching with you
16:49, why the tray date is June? It's May, isn't it?
these tools are gonna assist us in learning alot of stuff quickly, we will get university level education level through internet without paying a dime
And no jobs to use them for?
Im an elementary school teacher and I feel that my consistent use of AI has teleported me 5-10 years beyond most colleagues
I have created things this year that I never would have dreamed of - and it was all with less effort that I ever could have imagined. AI is truth.
@@Brax1982Why will we need jobs though? With AI, things will be dirt cheap. Imagine you don't need to buy food because you have your own personal robot growing it for you, it will build your car, your house, etc. It can work 24/7 and you can focus on doing what you want to do most and AI can help you reach those goals, like learning a language or playing an instrument.
@@blisphul8084 Yeah. Imagine. That YOU will have that.
@@rickeykoga2312 Can you describe how you actually use it? I can't find a real use for it.
It needs to see my screen and be able to operate my apps. I want to just direct it and have it model something in Maya for me or build a program while conversing with me in real time to better understand what I want. It will only be as good as the data you give it. At some point it will be so good, it will feel like a handicap to not use it.
Do i understand right that i would have to buy a new laptop (copilot+pc) in order to be able to do any of the things they’ve shown? Or will, some of, the possibilities also come to a surface laptop i bought two months ago?
Everything that doesn’t require that NPU should be available to you
@@MattVidProso for the voice feature we won't need to buy that PC?
Recall is basically rewind on macOS it seems. I wonder how much memory it’ll take up.
Dedicated 25 gigs. Normal use about 5 gigs per month.
How is Recall any different than the Rewind app for the Mac?
Nothing different
obvious difference aside, that being Windows vs Mac, I'd hope for their sake they make the UX at least slightly improved upon compared to Rewind, but at least not EVERYTHING is sent to OpenAI / MS (only supposedly "relevant" captures are sent to oAI/MS, when using Rewind's AI search)
@@2l2l2lalJT they ruthlessly stole the whole interface too, smh
Now...Recall is something I DON'T like about AI progression. I feel that will intrude tremendously on our daily lives and send that data to corporate assholes who will then exploit it. Or if you're researching stuff like how crimes are committed and such (for like a crime video game or movie type idea for example) and that data then gets sent to police and now you're on some kinda watch list for just doing your job. Yeah, I don't like the potential corruption involved in that one at all. Hell no to that one, I'll stick to my primitive searching if it means I can protect myself and my personal searching data. Or "Remember the password I used to log in to my bank?", "Can you show me the card numbers I put in when I ordered this item?"....That's like a hackers dream come true right there.
Now the translation feature, fantastic, excellent progress to a future of universal understanding of all languages regardless where you're at. There are other things that can be good about it but Recall is a huge red flag for anyone with a brain.
...you do realize your devices are currently always connected to the internet, and your "research on stuff like how crimes are committed", not to mention every bit of data you send in emails, Slack Messages, etc., are already subject to man-in-the-middle, right? If you don't trust these features to properly E2E encrypt your data, why would you trust the ones you use already?
If you don't want it, simple disable it. Problem solved!
That’s already the state of affairs only now it will benefit me.
No it's still bloat on standby for people who don't want it. @@kastamuyog
Google has had this kind of feature for almost 2 decades. But I think they turned off access to it when they shut down that G+ thing. But it took a lot of effort to keep it from knowing everything that happened on your computer, especially because I remember it wasn't too hard to dig up other people's information on that computer if you knew the right google-fu.
Hi matt, were u previously known as mattvideo productions? i remember i subscribed to u when u had a few hundread subs.
How do I access the 4o image creato? I have ChatGPT Plus, but it still uses Dalle even though I have 4o selected.
They are not rolling it out any time soon. It is only the model. Outputs are not multi-modal. The host is talking about the examples on the OpenAI page.
Very simple question. Can this "runs local" thing work without a net connection, YES OR NO?
Meaning GPT-4o? Good question, anyone have the answer?
nope ..
7:15 how does that even all fit into context length
I'm just wondering why the voice of the AI has some buzz sound ?
13:00 this is interesting, he can give advice on software
The Windows History Recall feature, which I believe is named "Recall," works by consistently or intermittently taking screenshot images in the background whenever any changes are made on the screen. This occurs as you work within Windows, while a custom Microsoft LLM simultaneously interprets all images and context to enable searching by phrase rather than requiring specifically related keywords.
I'm quite confident that it's not actually "GPT-4o" being used as the LLM powering this feature, as the host suggested, but rather a custom, fine-tuned version trained by Microsoft based on GPT-4.
It is supposed to run entirely on the edge, meaning within the PC hardware itself, without requiring an outside internet connection or reliance on Microsoft's "cloud AI" to function.
As of now, I have no way to verify whether this claim by Microsoft regarding Recall being entirely Edge-bound is true. Someone smarter than me will need to examine the code behind this feature once it's released to ascertain what truly occurs while this feature is active.
Seems pretty cool though and I believe could be useful!👍🏻
Extremely useful. Also..."Microsoft is entitled to up to 49% of the for-profit arm of OpenAI's profits" because they invested EARLY, and have invested 13 Billion dollars, so far.
Pretty intuitive stuff. I lose stuff all the time. This could help my scatter brain.
Yes i want it to not just watch what im doing for work but watch what im doing in life to help keep me on track and inspire me to exercise etc when i need too. i recently used the browser one to inspire me to exercise, but itd be cool if it was there like a robot with me and be like "hey u've been sitting here on X for a long time. maybe nows a good time to do a little exercise" or whatever that can help break me out of the trance i can sometimes get into
I never expected to meet Jarvis in my lifetime
Isn't Meta ahead in live translation AI?
Cant wait for June 18