AI Updates

News:

  • Large News for LLMs:

  • The coming awareness OpenAI and Meta, two of the biggest players in the AI game, are on the verge of releasing new models that promise to take AI capabilities to the next level. We're talking about ChatGPT-5 and Llama-3, and according to the chief researchers at these companies, these models will have enhanced reasoning and planning abilities that could be a real game-changer.

    So, what exactly does this mean? Well, Yann LeCun, Meta's chief AI scientist, emphasized that AI needs to evolve beyond just generating text. He believes that the future of AI lies in its ability to conduct thoughtful planning and reasoning, much like a human would.

    Meta is also focuses on developing models that can reason, plan, and possibly even retain memory. This is a big deal, folks. If AI can start to think and reason like a human, the possibilities are endless.

    Now, before we get too carried away, it's important to note that sentience, or true consciousness, in AI is still a distinct and unattained goal. We're not talking about AI that can feel emotions or have subjective experiences. But still, the advancements being made by OpenAI and Meta are a significant step towards what some might call 'superhuman cognition'.

    And here's the fascinating part: we might immediately see these new models in action. Both companies mentioned specific timelines, with OpenAI's GPT-5 expected in 2024 and Meta set to share early versions of Llama-3 in late 2024.

  • GPT trying to stay on top: If you're a regular ChatGPT user, you might have noticed that the usage limit policy has changed. Previously, the policy clearly stated that users had a limit of 40 messages every 3 hours. But now, the policy says, "Usage limits may apply." Pretty vague, right?

    Well, this change in wording reflects some actual changes in the usage limits themselves. Some users have reported that their limits have increased, allowing them to send more messages in a given period. That's excellent news for folks relying on ChatGPT for work or research.

    But here's where it gets a bit murky. Other users have revealed that OpenAI has introduced much stricter usage limitations. So, while some people enjoy more flexibility, others find themselves more restricted than before.

    Even more frustrating is that OpenAI has yet to make official announcements about these changes. They've quietly updated the policy wording but still need to provide clear information about the new limits.

    As a result, some users are starting to look for alternatives to ChatGPT. Platforms like Claude 3 and Gemini are seeing an uptick in interest as they offer greater flexibility and more transparent usage policies.

  • Meta is bringing the Llama. According to a leaked report from a Meta employee, the company is gearing up to launch two smaller versions of its highly anticipated Llama 3 large language model, or LLM, next week. This move will generate buzz and excitement ahead of the full Llama 3 launch.

    Now, what's the deal with these smaller LLMs? Well, the insider suggests they'll be more cost-effective and quicker at processing than the full-sized Llama 3. This makes them ideal for developers who want to build AI software for mobile devices, where efficiency and speed are key.

    But don't let their smaller size fool you. These mini Llamas are still expected to pack a severe punch in natural language processing and generation.

    Of course, the real star of the show will be Llama 3 itself. When it finally drops, this powerhouse LLM is expected to go head-to-head with OpenAI's GPT-4 in terms of its ability to understand, process, and generate large amounts of text and images.

    This is a big deal, folks. GPT-4 has been making waves in the AI world, and if Llama 3 can match or even exceed its capabilities, it could be a game-changer for Meta and the wider AI landscape.

  • EU vs Big Tech: The UK's Competition and Markets Authority, or CMA, has recently raised serious concerns about the concentration of power in the AI industry. They're worried that the deep integration of companies like Google, Amazon, Microsoft, Meta, and Apple across the AI sector could lead to unfair market practices and ultimately harm consumer interests.

    These tech giants have been instrumental in driving innovation and pushing the boundaries of what's possible with AI. Their partnerships and collaborations have led to some incredible breakthroughs. However, the CMA is concerned that this concentration of power could stifle competition and limit opportunities for smaller players to enter and thrive in the market.

    Imagine a world where a handful of companies control most AI development and deployment. They could dictate the direction of the industry, set prices, and limit consumer choice. That's a scary thought, especially considering how pervasive AI is in our daily lives.

    The CMA acknowledges the potential benefits of partnerships in the AI space, but it urges careful oversight to ensure that the market remains open and competitive. It wants to ensure that there's room for new entrants, innovative startups, and diverse perspectives in the AI landscape.

  • A new bill would force AI companies to reveal the use of copyrighted art https://www.theguardian.com/technology/2024/apr/09/artificial-intelligence-bill-copyright-art

  • Meta makes changes: If you recall, back in February, Meta faced criticism from its Oversight Board, which independently reviews the company's content moderation decisions. The board argued that Meta's existing content policies were narrow, incoherent, and confusing. Well, it seems like Meta has taken that feedback to heart.

    Starting in May, Meta will expand its use of the "Made with AI" label to a broader range of AI-generated content, including videos, images, and audio. The goal is to help users differentiate between genuine and deceptive content.

    Meta already uses an "Imagined with AI" label for photorealistic images created with its Meta AI feature. However, the company agrees with the Oversight Board that its approach needs to be narrower.

    Interestingly, Meta has also decided to stop removing AI content unless it violates other policies, like voter interference or bullying. Instead, they'll rely on labeling to avoid infringing on freedom of speech.

    It's a tricky balance, for sure. On the one hand, you want to protect users from deceptive content, but on the other, you don't want to stifle creativity or limit free expression. Meta is trying to strike that balance with these new changes.

    What do you think about this approach? Is labeling enough, or should there be stricter regulations around AI-generated content?

  • DALL-E Expands: OpenAI, the company behind the incredible DALL-E model, has just announced two new features to make creating and editing images more fun and intuitive.

    First up, let's talk about the new editing capabilities. You'll see an editing icon when DALL-E generates an image for you. This allows you to highlight a specific area of the image and type in a prompt to refine it. For example, you could select a car in the image and type "make this red" or "add racing stripes," DALL-E will update the image accordingly. Or, if you've generated a portrait, you could select the face and type "add glasses" to give your subject a new look.

    But that's not all! OpenAI also introduces preset-style prompt suggestions to help you get even more creative with your images. Let's say you've generated a nice landscape but want to see what it would look like in a different style. DALL-E will offer suggestions like "re-generate this image in a sci-fi style" or "give this image a painterly look." These prompts will help you explore new artistic directions and create stunning visuals.

  • DALL-E in the Military Microsoft, the tech giant that backs OpenAI, has reportedly pitched the company's text-to-image model, DALL-E, to the US Department of Defense, or DoD. The idea is that this AI could be used to train battle systems and assist military leaders in identifying targets.

    Before we jump to any conclusions, it's important to note that Microsoft has yet to sell any tools to the DoD. However, the Pentagon has clarified its mission to "accelerate the adoption of data, analytics, and AI across the DoD." So, it seems like it's only a matter of time before we see AI used in military operations.

    Interestingly, this pitch from Microsoft came shortly after OpenAI significantly changed its policy. Previously, the company had a line prohibiting its tools from being used for "military and warfare" purposes. However, that line was recently removed, and just days later, OpenAI began working with the DoD.

    This raises some serious ethical questions about the role of AI in warfare. On one hand, AI has the potential to make military operations more precise and efficient, potentially reducing casualties. But on the other hand, there are concerns about the accountability and transparency of using AI in such high-stakes situations.

  • Chips, Chips, and More Chips: Under the CHIPS and Science Act, legislation aimed at boosting domestic AI chip production, the US government has announced plans to award a whopping $11.6 billion in grants and loans to Taiwan Semiconductor Manufacturing Company, or TSMC for short. This funding is designed to help TSMC build three AI-chip production factories in Arizona.

    Now, this deal is pretty big. It's the most significant foreign direct investment in a brand-new development project in US history. As part of the agreement, TSMC will also increase its US investments from $40 billion to $65 billion. That's a lot of zeros!

    So, what does this mean in practical terms? Well, two of the factories are already under construction and expected to start production in 2025. The third factory, which is expected to produce advanced chips, has yet to start construction but is in the pipeline.

    But the impact of this investment goes beyond the factories themselves. TSMC's investment is expected to create over 6,000 high-tech manufacturing jobs and 20,000 construction jobs. To ensure that local workers have the skills to take advantage of these opportunities, TSMC also includes a $50 million training budget to upskill the workforce.

    This is a fascinating development for the US tech industry. By investing in domestic AI chip production, we're creating jobs, boosting the economy, and positioning ourselves as a leader in this critical technology.

    Of course, there are still challenges ahead. Building advanced chip factories is complex and expensive, and delays or setbacks are always risky. But overall, this is a positive step forward.

  • Even more Chips: Intel has just unveiled its latest and greatest AI chip, the Gaudi 3, which looks like it could be a serious contender against NVIDIA's H100 chip. NVIDIA currently dominates the market for powering AI applications at tech giants like Google and Microsoft.

    Now, here's where things get interesting. Intel claims that they've tested the Gaudi 3 on models like Meta's open-source Llama, and the results are pretty impressive. According to Intel, the Gaudi 3 is a whopping 40% more power-efficient, 50% faster, and uses less power than NVIDIA's H100 chips. If these claims hold up, it could be a game-changer for the AI industry.

    But the Gaudi 3 isn't just about raw performance. It also represents a strategic shift for major tech companies, or "hyper-scalers," as they're called in the industry. Companies like Microsoft, Google, and Meta rely heavily on NVIDIA's chips to power their AI systems. With the introduction of the Gaudi 3, these companies will have an alternative option, allowing them to reduce their dependence on a single supplier. This could be an intelligent move regarding business continuity and negotiating power.

    So, when can we expect to see the Gaudi 3 in action? Intel says it will be available in the third quarter of this year, and several companies, including Dell, have already confirmed that they plan to use these chips in their AI systems.

    Competition drives innovation, and with Intel entering the fray, we could see even more rapid advancements in AI hardware in the coming years.

  • Gmail Update: Google is on the verge of launching a game-changing feature called "summarize this email." As the name suggests, this new tool will give you concise summaries of individual emails, especially those lengthy and information-packed ones we all dread.

    Some might think, "Wait a minute, doesn't Google already have something like this?" And you'd be partially correct. Google offers a similar feature for paying Gemini Workspace users but has limitations. For one, it's only available on desktop. And secondly, it can only summarize entire email threads, not individual emails.

    But this new "summarize this email" feature is different. It'll appear as a handy button right under the subject line of an email, making it super easy to use. And the best part? It'll work hand-in-hand with Gmail's existing "reply suggestions" feature currently available on the Android app.

    Imagine understanding a complex email with just a click and having suggested replies ready to go. It's like having a personal email assistant!

  • Talk to me Gmail: If you're a Gmail user, you might be familiar with the "Help Me Write" feature, which generates draft emails based on text prompts. Well, get ready for an upgrade because Google has just announced that this feature is getting voice prompts! That's right, soon you'll be able to dictate your prompts instead of typing them out.

    Imagine this scenario: you're on the go, maybe driving or walking, and you remember that you need to send an important email. With this new voice prompt feature, you can speak your thoughts, and Gmail's AI will turn them into a structured, written draft. Talk about convenience!

    But wait, there's more! "Help Me Write" is also getting a new auto-refine feature. Once you have your draft, Gmail will automatically complete sentences, correct spellings, and fix any errors. It's like having your email editor ensure your messages are polished and professional before you hit send.

    Now, here's the catch: for now, these features are only available to Google Workspace users. But don't worry; the rumor mill suggests that Google plans to integrate these features into the regular version of Gmail within the following year. So, even if you're not a Workspace user, you might experience these AI-powered email enhancements quickly.

  • Can I get an A: The Texas Education Agency, or TEA, is currently testing an AI-powered scoring system to grade the State of Texas Assessments of Academic Readiness exams, more commonly known as STAAR. This move could have significant implications for both the agency and the students taking these exams.

    One of the most notable impacts of this test is the potential cost savings for the TEA. By using AI to grade exams, the agency plans to hire around 2,000 human graders this year, a significant decrease from the 6,000 graders they hired last year. This reduction in human graders could save the TEA a whopping $15 to $20 million annually.

    But how reliable is this AI scoring system? The AI was trained on 3,000 exam responses, which is a solid foundation. However, the TEA is taking some precautions to ensure accuracy. If an answer confuses the AI, perhaps because it includes slang or is worded unusually, it will be flagged for human review. Additionally, a quarter of all AI-graded results will be randomly selected for re-scoring by human graders.

    This hybrid approach of AI and human grading intelligently balances efficiency and accuracy. It allows the TEA to take advantage of AI's speed and cost-effectiveness while still maintaining human oversight to catch any potential errors.

  • AI makes Ads for Coke. Google and WPP, one of the world's largest advertising groups, have just signed a groundbreaking agreement. WPP, which creates ads for household names like Coca-Cola, L'Oréal, and Nestlé, is going to integrate Google's Gemini into its ad operations. This is a big deal, folks!

    So, what exactly is Gemini, and why is this partnership so significant? Well, Gemini is Google's powerful AI technology, and by combining it with WPP's existing AI platform, which is already used to create, test, and optimize ad campaigns, WPP is looking to supercharge its advertising effectiveness.

    Imagine this: you're scrolling through your social media feed and seeing an ad that seems tailored just for you. It speaks to your interests, lifestyle, and maybe even your current mood. That's the kind of hyper-targeted, super-relevant advertising that this Google-WPP collaboration could enable.

    By leveraging Gemini, WPP's clients will gain deeper insights into their audiences. They'll be able to predict which content will be most effective and continuously optimize their ad campaigns based on real-time data. It's like having a crystal ball that tells you exactly what your customers want to see!

  • Replace your iPhone? Humane, a company founded by former Apple employees, has officially launched their much-anticipated AI Pin, and it's got a pretty hefty price tag of $699. But before you balk at the cost, let's dive into what makes this device so unique.

    The AI Pin is designed to harness the power of generative AI and integrate various large language model-based systems to provide optimized functionality. In other words, this little device packs a serious punch regarding AI capabilities.

    But the AI Pin isn't just about showcasing the latest in AI tech. Humane has a broader mission: to create a range of devices that reduce our smartphone dependency. They want to help us break free from the constant pull of our phones and live more present, engaged lives. And the AI Pin is just the first step in this vision.

    Now, let's talk about what you get for that $699. The AI Pin package includes not only the device itself but also an extra battery and an AI charging case. And if you want to personalize your AI Pin, Humane offers color upgrades and accessories, although those will cost you extra.

    But the real kicker is the subscription service. After a three-month introductory period, users will need to pay $24 per month to continue accessing the AI Pin's full functionality. This subscription model is becoming more and more common in the tech world, and it's something to keep in mind when considering the long-term cost of the device.

    Creator Tools:

  • Claude: A conversational AI platform designed for nuanced and contextually aware interactions, aiming to provide human-like conversation experiences. https://claude.ai/chats

  • ChatGPT: A platform offering access to OpenAI's GPT model, tailored for engaging in conversational responses, providing information, and generating text-based content. https://chatgpt.com/

  • Pika Labs: A creative platform focused on AI-driven art creation, allowing users to explore and create digital artworks with the assistance of artificial intelligence. https://pika.art/

  • Runway: A creative toolkit powered by AI, enabling users to apply machine learning models to video, image, and text projects for innovative content creation. https://runwayml.com/

  • Leonardo: An AI platform (not widely known as of my last update, so this description is speculative) likely aimed at enhancing digital art creation or providing AI-based tools for creative processes. https://leonardo.ai/

  • Storyblocks: A stock media platform offering royalty-free videos, images, and audio clips for content creators to use in their projects. https://www.storyblocks.com/

  • Dalle-3 on Bing: Microsoft’s integration of OpenAI's DALL-E model into Bing allows users to generate unique images based on textual descriptions through a web interface. https://www.bing.com/images/create

  • Ideogram: An AI-driven platform focused on generating ideograms or visual symbols that represent ideas or concepts, facilitating creative visual communication. https://ideogram.ai/t/explore

  • MusicFX: A Google experiment that allows users to explore the creation of music using AI, part of Google's AI Test Kitchen, focusing on experimental AI applications in music. https://aitestkitchen.withgoogle.com

  • Mubert: An AI-powered platform for generating unique music streams, allowing creators to produce music through algorithmic composition. https://mubert.com/

  • Suno: An AI tool focused on enhancing voice communication by offering real-time, AI-driven voice analysis and improvement features (based on the description, this is speculative). https://www.suno.ai/

  • ElevenLabs: Offers AI technology for creating realistic voice synthesis and cloning, enabling high-quality voice generation for various applications. https://elevenlabs.io/

  • Adobe Speech Enhancer: A tool designed to improve the quality of audio recordings, especially in podcasting, by removing noise and enhancing speech clarity. https://podcast.adobe.com/enhance

  • Timebolt: A video editing software that automates the process of cutting out silences from video content, making content creation more efficient. https://www.timebolt.io

  • Descript: A multi-tool platform for audio and video editing that features transcription, text-to-speech, and media editing capabilities, aimed at content creators. https://www.descript.com/

  • HeyGen: Provides AI-driven tools for generating written content, aiming to assist in the creative writing process with the help of artificial intelligence. https://heygen.com/

  • Opus Clip: A platform (as of my last update, not widely recognized, so description is speculative) likely focused on video editing or creation tools enhanced by AI. https://www.opus.pro/

  • TubeBuddy: A browser extension and mobile app designed to help YouTube creators optimize their videos, manage their channels, and grow their audience. https://futuretools.link/tubebuddy-com

  • Invideo: An online video creation platform offering tools and templates for creating professional-quality videos for marketing, social media, and more. https://invideo.io/

  • LTX Studio: Specializes in leveraging AI for text-to-video generation, allowing users to create video content from written narratives. https://ltx.studio/