Artificial Intelligence - Page 32

AI music generators have become all the rage since the explosion in popularity of AI-generation tools, but now we are starting to hear the oddities that can come out of AI generation.

A Reddit user has posted a short clip that was created using the music generation software known as Suno. The user wrote that with the 24-second clip, the AI sounded like it was crying at the end of the video and that this crying wasn't part of the prompt that created the clip. The user wasn't alone, as another commented that an AI-generated song that has since been posted on Spotify features an outburst at the end where users can audibly hear "No!".

So, what could be causing these strange and seemingly emotional outbursts? Well, these AI systems are designed to create music based on the keywords provided in the user's prompt. It appears that Suno is attempting to create a human-esc outro, which can often feature fading pieces of audio and sometimes single words or sounds.

YouTube has announced it's working on a new set of tools designed to detect AI-generated content across its platform, and that these tools have been created with the intention of protecting the likeness of creators.

In a new post on the official YouTube blog, the company's Vice President of Creator Productors at YouTube, Amjad Hanif, explained how the new tools represent YouTube's commitment to "responsible AI development," and part of that is regulating the content on its platform that AI is helping create. The blog post reveals the video platform has created a new tool that is capable of detecting the signing voice of a musician, or the musician's "likeness".

The same principle has been applied to another tool that's designed to identify AI-generated content showing the faces of actors, creators, athletes, political figures, and more. The tools are meant to be guardrails for how YouTube is going to deal with the influx of AI-generated video content on its platform when AI-generation tools eventually make it into more people's daily devices. That isn't to say AI-generated content isn't already being posted on YouTube because it certainly is, and at an increasing rate.

Samsung has announced it is partnering with TSMC on co-developing bufferless HBM4 memory chips for future AI chips at the SEMICON Taiwan 2024 forum on Thursday.

Samsung is the world's largest memory chipmaker, partnering with TSMC, the world's largest contract chip manufacturer, with the South Korean and Taiwan semiconductor giants working together on bufferless HBM4 memory in order to strengthen their positions in the constantly-evoling AI chip market.

Dan Kochpatcharin, the head of Ecosystem and Alliance Management at TSMC, said during SEMICON Taiwan 2024 that the two companies were developing a bufferless HBM4 chip. Samsung makes its own HBM4, with TSMC forming a "triangular alliance" with SK hynix and NVIDIA on future HBM and AI designs. SK hynix is second to Samsung (and also native to South Korea) but this new development between Samsung + TSMC is very interesting.

Chinese cloud service providers have "renting services" where they rent out their hardware stack, with prices that are radically cheaper than those available in the United States... as low as $6 per hour.

The Financial Times reports that Chinese CSPs (cloud service providers) are renting out their AI GPU hardware, with small Chinese CSPs providing companies with an AI server that packs 8 x NVIDIA A100 AI GPUs that costs around $6 per hour to rent them out. If you're in the US, that would cost you around 50% more at $10 per hour.

US sanctions to the side, NVIDIA's newer H100 and older A100 AI GPUs are easily available in China, which is why the rental costs are so much lower than other regions. It's estimated that over 100,000 x NVIDIA H100 AI GPUs are in China right now, and they're openly being sold on Chinese marketplaces, smuggled all across the country according to one Chinese startup founder.

Micron has just announced its shipping production-capable HBM3E 12-Hi memory in up to 36GB capacities, pushing 1.2TB/sec of memory bandwidth ready for AI GPUs.

The new Micron HBM3E 12-Hi features an impressive 36GB capacity, which is a 50% increase over current HBM3E 8-Hi stacks, allowing far larger AI models like Llama 2 with 70 billion parameters to run on a single AI processor. This increased capacity to 36GB allows faster time to insight by avoiding CPU offload and GPU-GPU communication delays.,

Micron's new HBM3E 12-Hi 36GB delivers "significantly" lower power consumption than competitors' HBM3E 8-Hi 24GB memory, with Micron's new HBM3E 12-Hi memory pushing over 1.2TB/sec of memory bandwidth at a pin speed of over 9.2Gbps. These combined benefits of Micron's new HBM3E memory modules over maximum throughput with the lowest power consumption, ensuring optimal outcomes for power-hungry data centers of the future.

TIME has released its list of what it says are the top 100 most influential people in AI, a list that anyone that knows anything about AI would expect to see SpaceX, Tesla and xAI boss Elon Musk on the list... but nope. TIME doesn't think he's influential enough, but actress Scarlett Johansson is.

In the cover, you can see NVIDIA CEO Jensen Huang, AMD CEO Dr. Lisa Su, and even YouTuber "MKBHD" who, as much as I somewhat enjoy some of his videos, is not top 100 most influential in AI material, sorry Marques. Elon Musk, however, has the largest AI GPU cluster in the world, millions of Tesla vehicles on the road capturing video and information and feeding it into an AI, X is feeding billions of bytes of data per day into AI, and he has self-landing rockets with SpaceX.

I mean, TIME isn't just out of touch -- this is showing their bias. Rewinding the clock back to 2016-2020, when anything Trump did saw people who didn't like Trump referred to as "TDS" or Trump Derangement Syndrome. This is some serious EDS, or Elon Derangement Syndrome. Just because they don't like his politics -- he's voting Trump, and is fully 100% behind President Trump -- so they omit him from a list where he should've been #1. Right.

SK hynix will begin mass production of its new 12-layer HBM3E memory chips before the end of September, with shipments of the new AI memory expected in Q4 2024... not long from now.

The South Korean memory leader isn't just teasing its 12-layer HBM3E memory, but its next-gen HBM4 memory that will be shipping in 2H 2025 ready for NVIDIA's next-generation Rubin R100 AI GPU which uses the ultra-fast next-generation HBM4 memory standard.

The new comments about 12-layer HBM3E and future-gen HBM4 AI memory were made during SEMICON Taiwan 2024, of which SK hynix was in attendance. SK hynix showed off its new 12-layer HBM3E AI memory at FMS 2024 back in July, but it appears we're even closer to it becoming a reality.

Before launching its new Copilot+ PC range, Microsoft pulled one of its flagship Windows 11 AI features, 'Recall.' The search tool leveraged the Copilot+ NPU to search your PC history to 'recall' or find things you've looked at or worked on. An example could be asking Recall to find that funny YouTube video you watched three days ago.

The only problem was that Recall worked by constantly taking screenshots of your desktop without any security or filters. From there, the AI builds a searchable index with everything categorized without regard for privacy.

Once this became known, the controversial AI feature made headlines, and it continued to make headlines until Microsoft pulled it from the Copilot+ PC launch to retool its functionality and security features. Well, it's coming back, and you won't be able to uninstall Recall once it does.

A few select companies in the technology sector are leading the exponentially growing push into artificial intelligence (AI) development, and Elon Musk has seemingly pushed ahead of everyone by switching on what he's calling the world's most powerful AI training system.

According to Musk, the team at xAI began working on the AI training system 122 days ago, and they were able to stand it up completely in that brief amount of time. The new system consists of 100,000 NVIDIA H100 GPUs and has been named "Colossus". Additionally, Musk revealed that Colossus isn't completely finished, as it's expected to gain another 100,00 NVIDIA H100 GPUs over the "next few months".

Musk's claim that xAI has built the "most powerful" AI training system is based on the number of GPUs or training horsepower the system has available to it, which is estimated to be more than any other system publicly revealed. Musk's new AI training model will likely be used to train X's large language model, which powers the social media platform's AI chatbot, Grok.

OpenAI is close to releasing new AI models that will be 100x more powerful than GPT-4, with OpenAI Japan teasing GPT-Next, and it's 100x more powerful than GPT-4 without wasting more computing resources.

The new codenamed "Strawberry" AI model was teased alongside "Orion", with Strawberry ending up as GPT-Next and expected by the end of the year, while Orion will be unleashed in 2025. OpenAI's new GPT-Next aka Strawberry has something researchers call "System 2 thinking".

What is System 2 thinking? This will allow GPT-Next to take the time to deliberate and reason through problems, versus just predicting longer and longer sets of tokens to complete its responses. System 2 thinking has impressive results: scoring over 90% on the MATH benchmark, a collection of advanced mathematical problems, reports Reuters.