Category: [Image, Video, Article, Tweet, Thread]
Hook:
- a
- b
- c
- d
Tweet Summary:
Category: [Image, Video, Article, Tweet, Thread]
Hook:
- a
- b
- c
- d
Tweet Summary:
Category: [Image, Video, Article, Tweet, Thread]
Hook: AI That Can Play MINECRAFT?
- A new study introduces Voyager, an LLM-powered agent in Minecraft, which can learn and evolve without any human intervention.
- Voyager explores the world, builds new skills, and makes original discoveries independently. It interacts with GPT-4 via blackbox queries, bypassing the need for model parameter fine-tuning. It's 3.3x more efficient in obtaining unique elements, travels 2.3x longer distances, and progresses up to 15.3x faster in the game compared to prior SOTA!
- Voyager continuously improves itself by writing, refining, committing, and retrieving *code* from a skill library.
- A major leap towards developing autonomous agents for complex, open-ended worlds. New standard for player-less game-testing. Increment our understanding of the potential of LLM in embodied tasks.
- You can see AI in action: https://twitter.com/DrJimFan/status/1662115266933972993
- The code is completely OPEN SOURCE. Voyager is currently text-only, but can be augmented by visual perception in the future.
- Generally capable, autonomous agents are the next frontier of AI. They continuously explore, plan, and develop new skills in open-ended worlds, driven by survival & curiosity.
- Remember when we were amazed by AI beating human at chess? Now it's teaching itself Minecraft!
Script: Today, I have an exciting topic to discuss. Have you ever wondered if artificial intelligence could play Minecraft? Well, hold onto your seats because a groundbreaking study has introduced an incredible agent called Voyager, powered by the latest language model, GPT-4.
This study showcases Voyager's remarkable abilities in the Minecraft world. Unlike previous models that required human intervention, Voyager can learn and evolve on its own. It explores the game, develops new skills, and makes original discoveries autonomously. What's even more impressive is that Voyager interacts with GPT-4 through blackbox queries, eliminating the need for model parameter fine-tuning.
The results are astonishing. Voyager outperforms state-of-the-art models by being 3.3 times more efficient in obtaining unique elements, traveling 2.3 times longer distances, and progressing up to 15.3 times faster in the game. It achieves these incredible feats while continuously improving itself by writing, refining, committing, and retrieving code from a skill library.
This advancement is a significant leap towards developing autonomous agents for complex, open-ended worlds. It sets a new standard for player-less game-testing and expands our understanding of the potential of language models in embodied tasks. Let’s check out the actual video to witness Voyager's impressive gameplay. https://twitter.com/DrJimFan/status/1662115266933972993
What's even more exciting is that the code behind Voyager is completely open source. Although it currently operates in a text-only format, there are possibilities of incorporating visual perception in the future. Autonomous agents like Voyager represent the next frontier of AI, continuously exploring, planning, and developing new skills in open-ended worlds driven by survival and curiosity.
Just think about it: we were amazed when AI defeated humans in chess, and now it's teaching itself how to conquer Minecraft.
Category: [Image, Video, Article, Tweet, Thread]
Hook: Text To 3D Just Improved Dramatically
- This croissant is generated by a text-to-3d AI model. We've hit the pinnacle of bread-tech. https://twitter.com/ben_ferns/status/1662051566973763585?s=20
- It’s using a process called Variational Score Distillation (VSD). It’s an improvement over the existing Score distillation sampling (SDS) which suffers from over-saturation, over-smoothing, and low-diversity problems. VSD aims to fix these problems
- This approach, dubbed ProlificDreamer, can generate high rendering resolution (i.e., 512 × 512) and high-fidelity NeRF with rich structure and complex effects (e.g., smoke and drops). Further, initialized from NeRF, meshes fine-tuned by VSD are meticulously detailed and photo-realistic.
- You can see the full resolution image of 3D generated croissant here: https://twitter.com/ben_ferns/status/1662074088658206720/photo/1
- You can see more text to 3D models https://twitter.com/ben_ferns/status/1662052238523785219?s=20
Script: Get ready to be amazed because text-to-3D technology just took a giant leap forward.
Picture this: a croissant generated by an AI model using text-to-3D technology. You have to see it to believe it. https://twitter.com/ben_ferns/status/1662051566973763585?s=20 https://twitter.com/ben_ferns/status/1662074088658206720/photo/1
This groundbreaking achievement is made possible by a new process called Variational Score Distillation (VSD). It's a significant improvement over the existing Score Distillation Sampling (SDS) method, which had some drawbacks like over-saturation, over-smoothing, and low diversity. VSD aims to address these issues and take text-to-3D to new heights.
The project is dubbed ProlificDreamer. With ProlificDreamer, the AI can generate high-resolution renderings, reaching an impressive 512 × 512 pixels. Not only that, but the resulting 3D models exhibit rich structure and complex effects like smoke and drops. The level of detail and realism is truly remarkable.
This technology doesn't stop at croissants. There are other incredible text-to-3D models out there. As you can see from this video here https://twitter.com/ben_ferns/status/1662052238523785219?s=20
Innovation never ceases to amaze us, and this advancement in text-to-3D technology is proof of that.
Category: [Image, Video, Article, Tweet, Thread]
Hook: Make Photoshop do MAGICAL Things in Seconds.
- I’ve come across incredible use cases for Photoshop’s new Generative Fill feature.
- You can now edit with words. This will take a lot of hard work and save lots of time. YOu can see in this video how he places steak on his plate using just word prompts. Also he changed the color of fork from silver to gold. https://twitter.com/heyBarsee/status/1662113151096168450?s=20
- It can effortlessly combine two images. https://twitter.com/shota7180/status/1661874813177311232
- Fixing the Pope's viral photo with words is the real deal. This user added legs to the famous pope Drip photo. https://twitter.com/mrgreen/status/1661631011740950530?s=20
- Changing clothes on your picture doesn’t require a $200 course anymore. https://twitter.com/vw_artist/status/1661859844511473665?s=20
- Comparison of new Generative AI VS old Content-Aware Fill https://twitter.com/heyBarsee/status/1662113167122522114?s=20
- Adobe also added a new Remove Tool to get rid of unwanted images or objects. https://twitter.com/heyBarsee/status/1662113177755156480?s=20
Script:
Have you heard about Photoshop's new Generative Fill feature? Well, let me tell you, it's a game-changer!
Imagine being able to edit images with just words. Sounds incredible, right? Trust me, it's real, and it's going to save you a ton of time. In fact, I’vefound some mind-blowing examples that showcase the power of this feature.
Take a look at this video, where a user effortlessly places a realistic-looking steak on his plate using nothing but word prompts. And that's not all – he even changes the color of a fork from silver to gold. Unbelievable! https://twitter.com/heyBarsee/status/1662113151096168450?s=20
Generative Fill can seamlessly combine two images, creating stunning compositions in seconds. Just check out this tweet showcasing the impressive results. It's like magic unfolding before your eyes. https://twitter.com/shota7180/status/1661874813177311232
Now, prepare to have your mind blown once again. Remember that viral photo of the Pope? Well, someone actually used words to add legs to the famous Pope Drip photo. https://twitter.com/mrgreen/status/1661631011740950530?s=20
And here's the best part – you no longer need to invest in expensive courses to change clothes in your pictures. With Generative Fill, it's as simple as a few clicks. Check out this video. https://twitter.com/vw_artist/status/1661859844511473665?s=20
And if that's not enough, Adobe has also introduced a new Remove Tool, allowing you to effortlessly get rid of unwanted images or objects. https://twitter.com/heyBarsee/status/1662113177755156480?s=20
So, whether you're a professional designer or just someone who loves playing around with photos, Photoshop's Generative Fill is here to make your editing dreams come true. Say goodbye to hours of manual work and hello to a world where Photoshop does magical things in seconds!
Category: [Image, Video, Article, Tweet, Thread]
Hook: Elon Musk Will Chip EVERYONE?!
- Elon Musk’s brain chip company, Neuralink gets FDA approval to launch its human clinical study.
- Neuralink hopes the implant it is developing will help paralysed people walk again - and cure other neurological issues.
- Here is a video of a monkey with Neuralink installed playing Pong with just its mind: https://twitter.com/mrgreen/status/1661981443994578945?s=20
- Blackrock Neurotech, a subsidiary of the investment management corporation Blackrock, has developed brain-computer interfaces (BCIs) that have already been successfully implanted in over 50 people.
- Is this something humanity is ready for? It's imperative we proceed with caution and carefully consider the full potential of risks and benefits of this technology as it becomes more available.
- You can join the waitlist here https://neuralink.com/patient-registry/ …. but I’ll pass though
Script:
I have some groundbreaking news that might make your jaw drop. Elon Musk's brain chip company, Neuralink, has received FDA approval to launch its human clinical study. That's right, we're talking about brain-computer interfaces!
Neuralink aims to revolutionize the way we approach neurological issues, and even enable paralyzed individuals to walk again. Imagine the possibilities!
Need proof? Just take a look at this incredible video of a monkey with Neuralink installed, playing Pong using only its mind. It's mind-blowing! https://twitter.com/mrgreen/status/1661981443994578945?s=20
Blackrock Neurotech, a subsidiary of the renowned investment management corporation Blackrock, has already successfully implanted brain-computer interfaces in over 50 people. The progress in this field is truly remarkable. Seems that Neuralink has some catching up to do.
However, let's not forget the importance of caution. As this technology becomes more available, we must carefully consider the full potential of both the risks and benefits it brings. Is humanity truly ready for such a leap?
Well, if you're feeling adventurous, you can even join the waitlist for Neuralink's patient registry. But personally, I'll pass. Let's see how this unfolds.
Category: [Image, Video, Article, Tweet, Thread]
Hook: TikTok Jumps the AI Train?
- TikTok said it is testing an AI chatbot called 'Tako' to help users discover content. The social media platform has set a disclaimer saying Tako is experimental and its responses could be inaccurate, warning users not to share private information with it
- Once available, users can access Tako by tapping a new button right above the profile icon. “If I’m watching food videos and ask for a recipe, I’ll get related TikTok videos for the recipe, or if I ask for good art exhibitions in Paris, it’ll show videos alongside a list of suggestions in the answer,” explained one journalist who had early access to Tako.
- One of the screenshots shows Tako suggesting the user to ask, “What is the significance of King Charles III’s coronation?” In practice, using Tako feels like talking to someone in a chat room, just like ChatGPT. However, it’s unclear whether Tako is based on a proprietary AI model from TikTok or a third-party one. https://9to5mac.com/wp-content/uploads/sites/6/2023/05/TikTok_Tako_test.jpg?resize=2048,1914
- The bot is being tested in global markets but is not yet available in the US.
- The platform already has a really smart algorithm for suggesting videos to users. Combined with AI, the suggestions can become even more precise. But more than that, as more people – especially teenagers – are spending more time on TikTok, having a chatbot could make it the main search engine for many internet users.
Script:
Exciting news from TikTok. It seems like the popular social media platform is jumping on the AI train with its latest experiment. They're testing out an AI chatbot called 'Tako' to help users discover content. But there's a catch, so make sure you stick around to find out more!
According to TikTok, Tako is still in the experimental phase, so its responses might not always be accurate. In fact, they've even issued a disclaimer, warning users not to share private information with the chatbot.
Once Tako becomes available, users will be able to access it by simply tapping on a new button located right above their profile icon. Early access journalists who got a sneak peek into Tako's capabilities have shared their experiences. For example, if you're watching food videos and ask for a recipe, Tako will provide you with related TikTok videos showcasing the recipe. Similarly, if you're curious about good art exhibitions in Paris, Tako will show you videos along with a list of suggestions. It's like having a chat room companion for all your queries!
However, it's still unclear whether Tako is based on TikTok's proprietary AI model or if they're using a third-party one. The platform is currently testing the chatbot in global markets but hasn't rolled it out in the US just yet.
TikTok already has a smart algorithm for suggesting videos, and now with the integration of AI through Tako, these suggestions can become even more precise. Moreover, as more and more people, especially teenagers, spend a significant amount of time on TikTok, having a chatbot like Tako could potentially turn TikTok into the go-to search engine for many internet users.
So, what do you think about TikTok's new AI chatbot experiment? Are you excited to try it out once it becomes available in your region?
Category: [Image, Video, Article, Tweet, Thread]
Hook: The Future of Podcasts?!
- AI conversations of Bill Gates and Socrates
- This is a tech demo using D-ID. It’s a web app that uses real-time face animation and advanced text-to-speech to create an immersive and human-like conversational AI experience. The free app lets you speak face-to-face with ChatGPT.
- In my opinion, it would have been even more realistic with Google SoundStorm
- I think very realistic conversations like this will be bread and butter in any streaming app.
- Just imagine yourself having a conversation with any historical figure and they are able to respond to your questions in their style.
Script: Imagine listening to conversations of historical figures like Bill Gates and Socrates. Well, thanks to an incredible tech demo, that's now possible.
Let’s take listen to this funny conversation about Bill Gates explaining Laptops to Socrates. It seems that socrates is familiar with MacBook but has never heard about Microsoft Surface.
This conversation was generated by a Web app called D-ID. It combines real-time face animation and advanced text-to-speech to create an immersive and human-like conversational AI experience. And the best part? It's absolutely free!
Now, I personally believe that incorporating Google SoundStorm would make it even more realistic. But let me tell you, the conversations you can have using this technology are mind-boggling. Just imagine talking to historical figures and having them respond to your questions in their own unique style.
I can't help but think that this kind of realistic interaction will become a staple in streaming apps. It adds a whole new level of immersion and engagement. And not just for historical figures, but for any topic or personality you can think of!
The future of podcasts is truly evolving, and technologies like D-ID are pushing the boundaries of what's possible. So, what are your thoughts on this? Can you picture yourself having these incredible conversations with historical figures?
Category: [Image, Video, Article, Tweet, Thread]
Hook:
- a
- b
- c
- d
Tweet Summary: