?? Welcome to AI Insights Unleashed! ?? - Vol. 29

Embark on a journey into the dynamic world of artificial intelligence where innovation knows no bounds. This newsletter is your passport to cutting-edge AI insights, thought-provoking discussions, and actionable strategies.

?? What's New This Week ??

OpenAI prepares ‘Project Strawberry’

OpenAI researchers are preparing to launch a new AI model, code-named Strawberry (previously Q*), that demonstrates superior reasoning capabilities in solving complex problems.

• Project Strawberry could be integrated into ChatGPT as soon as this fall, marking a significant leap in AI intelligence.
• Given extra “thinking” time, Strawberry can tackle subjective topics and solve complex puzzles like the New York Times Connections.
• OpenAI is using Strawberry to generate high-quality training data for another secretive upcoming LLM, reportedly code-named Orion.
• The new AI model could enhance OpenAI’s development of AI agents, potentially automating multi-step tasks more effectively.

If Strawberry lives up to the leaks, it could mark a significant leap in AI reasoning capabilities, potentially advancing OpenAI towards Stage 2 of its five-level?roadmap?to AGI. With ChatGPT reported to gain these capabilities this fall, we’re likely on the verge of seeing the next major wave of AI disruption.

China is coming for Tesla Optimus

At the 2024 World Robot Conference in Beijing, Chinese companies?showcased?27 humanoid robots alongside Tesla’s Optimus, signalling China’s ambition to dominate the industry.

• Chinese tech firms unveiled 27 humanoid robots at the expo, with Tesla’s Optimus being the only foreign competitor present.
• AGIBOT, founded by a Huawei alumnus, presented robots powered by large language models (LLMs) for industrial use and customer service.
• Other notable entries included Astribot’s S1 robot assistant capable of writing calligraphy and playing musical instruments, and Galbot’s wheeled robots for food delivery and retail tasks.
• Despite the impressive showcase, experts note that technological hurdles and high costs still create challenges for Chinese manufacturers.

China may be slightly behind in the AI race against the U.S., but it’s clear the country is committed to dominating the humanoid robotics race. With a whopping 27 China-based humanoid robots demonstrating a wide-range of use cases at the event, commercially available humanoids may be coming sooner than most expect.

Apple working on next-gen humanlike AI ‘personality'

Apple is reportedly developing an AI personality entirely separate from Siri for future Apple robotics devices. The company's first robotics projects are not expected to debut until 2026 or 2027. Apple will ship a brand new Apple Intelligence-fused Siri in iOS 18.1 later this fall. However, many of Siri's most powerful new capabilities aren't scheduled to arrive until sometime in 2025.

OpenAI and Anthropic partner with US gov

OpenAI and Anthropic just?signed?a groundbreaking agreement with the U.S. Artificial Intelligence Safety Institute to allow government access and testing of their AI models before public release.

• The U.S. AI Safety Institute will have access to major new models from both companies prior to and after their public release.
• This collaboration is a step toward AI regulation and safety efforts, with the U.S. government evaluating AI models’ capabilities and associated risks.
• The institute will provide feedback to OpenAI and Anthropic on potential safety improvements that should be made.
• These agreements come as AI companies face increasing regulatory scrutiny, with California legislators recently passing a broad AI regulation bill earlier today.

The two most popular AI companies in the world are granting the U.S. government access to unreleased models before release. This could reshape how AI is developed, tested, and deployed worldwide, with major implications around innovation, safety, and international competition in the AI space, for better or worse.

AI Inference Competition Heats Up

The competition for AI inference is heating up, particularly in terms of power efficiency. ML Commons has released the results of its latest AI inferencing competition. The round included first-time submissions from teams using AMD Instinct accelerators, the latest Google Trillium accelerators, chips from Untether AI, and Nvidia's new Blackwell chip. This article looks at the results - while Nvidia's chips are still ahead in many categories, chips from other manufacturers are starting to catch up.

AI startup reaches 100M token context

Magic just?developed?LTM-2-mini, a model capable of processing 100 million tokens of context — equivalent to about 10 million lines of code or 750 novels — and partnered with Google Cloud to build advanced AI supercomputers.

• LTM-2-mini can process and understand 100 million tokens of context given during inference, surpassing current models by 50x.
• The model’s innovative algorithm processes long sequences of data 1000x more efficiently than the current top-performing AI models.
• Magic is also partnering with Google Cloud to build supercomputers powered by Nvidia’s newest and most advanced GPUs.
• The company has raised more than $450 million in total funding, including a recent $320 million investment round.

This breakthrough in context length allows AI agents to process and reason over dense and complicated codebases, vast databases, and years of conversation history in a single inference. It’s a significant step toward creating AI assistants with near-perfect recall and memory.

OpenAI raises at $100B valuation

OpenAI is?reportedly?in talks to raise a new funding round at a valuation exceeding $100 billion, led by Thrive Capital, with Microsoft also expected to participate.

• The potential valuation of over $100 billion would be significantly higher than OpenAI’s previous $86 billion valuation.
• Thrive Capital is expected to invest around $1 billion in this round.
• OpenAI’s annualized revenue reportedly surpassed $3.4 billion earlier this year.
• The company is still, however, projected to lose nearly $5 billion by the end of the year and has already spent $8.5 billion on AI training and staffing.

Building AI is expensive, and raising billions of dollars at a $100B+ valuation would silence OpenAI’s critics who insist that the company is on its downfall. The increased valuation also suggests that the company has potential hidden breakthroughs behind the scenes, such as?Project Strawberry?and Orion.

ChatGPT teams up with ASU

OpenAI’s ChatGPT is?headed?to Arizona State University (ASU), where the university is integrating the AI assistant into over 200 projects across teaching, research, and operations.

• ASU is using ChatGPT Edu, a version designed for universities with enhanced privacy and security features.
• The university also launched an ‘AI Innovation Challenge’ for faculty and staff, receiving an overwhelming demand for using ChatGPT to maximize teaching, research, and ops.
• Key projects include an AI writing companion for scholarly work, 'Sam' (a chatbot for med students to practice patient interactions), and AI-assisted research recruitment.
• The partnership has inspired other institutions like Oxford and Wharton to pursue similar collaborations.

While some schools are attempting to resist AI, ASU is embracing ChatGPT to make learning more personalized and to prepare students for an increasingly AI-driven job market. As education continues to change in the age of AI, case studies like this will be instrumental in shaping the future of academia.

D-ID launches an AI video translation tool that includes voice cloning and lip sync

D-ID has launched an AI Video Translate feature that clones the speaker's voice and syncs lip movements in translated videos. Supporting 30 languages, it seeks to reduce localization costs for global campaigns. It is available to subscribers, with plans starting at $56 per year. The technology competes with similar offerings from companies like YouTube and Vimeo, as well as numerous AI voice cloning tools.

?? Key Developments ??

China’s AI Price War May Spark Global Tech Showdown

Chinese tech giants are slashing prices on AI models, creating intense competition and impacting the global AI market, including Silicon Valley. Companies like Baidu, Tencent, and iFlytek have drastically reduced prices, making advanced AI tools more accessible worldwide. This price war could democratize AI, allowing small startups and traditional businesses to compete with larger tech firms by leveraging cutting-edge technology.

However, the aggressive pricing strategy poses risks, such as reduced profitability for major players and potential security and privacy challenges for consumers. As AI becomes more commoditized, businesses and consumers could benefit from a broader range of AI-powered applications, but they must remain vigilant about data privacy and the authenticity of online information.

China’s new Qwen2 beats GPT-4o

Alibaba just?unveiled?Qwen2-VL, a new vision-language AI model that outperforms GPT-4o in several benchmarks — particularly excelling in document comprehension and multilingual text-image understanding.

• Qwen2-VL can understand images of various resolutions and ratios, as well as videos over 20 minutes long.
• The model excels particularly at complex tasks such as college-level problem-solving, mathematical reasoning, and document analysis.
• It also supports multilingual text understanding in images, including most European languages, Japanese, Korean, Arabic, and Vietnamese.

There’s yet another new contender in the state-of-the-art AI model arena, and it comes from China’s Alibaba. Qwen2-VL’s ability to understand diverse visual inputs and multilingual requests could lead to more sophisticated, globally accessible AI applications.

AI creates DOOM video game in real-time

Google researchers just?developed?GameNGen, an AI system that can simulate the classic game DOOM in real-time, running at over 20 frames per second and producing visuals nearly indistinguishable from the original game.

• GameNGen produces playable gameplay at 20 frames per second on a single chip, with each frame predicted by a diffusion model.
• The AI was trained on 900M frames of gameplay data, resulting in 3-second clips almost indistinguishable from the actual game by playtesters.
• Running on a single TPU, GameNGen handles Doom's 3D environments and fast-paced action without traditional game engine components.
• In tests, human raters could barely distinguish between?short clips?of the AI simulation and the actual game.

GameNGen is the first AI model that can generate a complex and playable video game in real-time without any underlying real game engine.?We’re at the fascinating time where soon, AI will be able to?create entire games on the fly, personalized to each player.

Discovering new drugs with AI

Scientists from China and the U.S. just?developed?ActFound, a new AI model that outperforms existing methods in predicting drug bioactivity, potentially accelerating and reducing costs in drug development.

• ActFound combines meta-learning and pairwise learning to overcome common limitations in AI drug discovery, like small datasets and incompatible measurements.
• The model was trained on 35,000+ assays (metal ore breakdowns) and 1.6 million experimentally measured bioactivities from a popular chemical database.
• In tests, ActFound outperformed nine competing models and showed strong performance in predicting cancer drug bioactivity.

ActFound could significantly speed up drug development by accurately predicting compound properties with less data and lower costs than traditional methods. While still in early stages, AI breakthroughs like this are the lesser-talked about developments that could end up saving millions of lives.

New AI breakthrough in solar tech

Researchers developed?a new AI-based method called NES-VMC that can accurately calculate the excited states of atoms and molecules, a challenge in physics and chemistry that previously delayed improvements in solar tech.

• NES-VMC (natural excited states variational Monte Carlo) accurately predicted quantum excited states on systems ranging from single atoms to benzene-sized molecules.
• The method outperforms leading computational chemistry techniques, often achieving chemical accuracy.
• Excited states are crucial for understanding light-matter interactions, key to improving solar cells, LEDs, lasers, and more.
• NES-VMC overcomes long-standing challenges in physics and chemistry that have hindered progress in these fields.

This AI-driven breakthrough could lead to more efficient solar cells, brighter LEDs, and more powerful lasers. The ripple effects could be dramatic: lower electricity costs, improvements in phone and laptop battery life and displays, faster fiber-optic internet, and so much more.

AI learns to plan better without humans

IBM Research and Cornell University recently?created?AutoToS, a system that teaches AI to solve complex planning problems at 100% accuracy — without needing a human to check its work.

• AutoToS is like a smart tutor for AI, helping it learn how to break down and solve tricky problems step-by-step.
• The system uses clever tests to check the AI’s work, pointing out mistakes and showing examples of how to do better without human interferance.
• AutoToS succeeded in teaching AI to solve complex puzzles, including classic problems like arranging blocks and solving Sokoban, a box-pushing game.

Right now, it’s difficult to trust AI agents to completely autonomously perform actions on your behalf, but AutoToS is solving complex tasks at a 100% accuracy. If this system works in the real world, it’s the next big step in creating more reliable AI assistants.

AI can 3D print lifelike human organs

Researchers at Washington State University recently?developed?an AI technique called Bayesian Optimization that dramatically improves the speed and efficiency of 3D printing lifelike human organs.

• The AI balances geometric precision, density, and printing time to create organ models that look and feel authentic.
• In tests, it printed 60 continually improving versions of kidney and prostate organ models.
• This approach significantly reduces the time and materials needed to find optimal 3D printing settings for complex objects.
• The technology also has potential applications beyond medicine — for example, in the computer science, automotive, and aviation industries.

With cheaper, lifelike 3D-printed human organs, medical students could better practice for surgery before operating on actual patients. Beyond medicine, this AI technique could help reduce manufacturing costs for a variety of things like smartphones, car parts, and even airplane components.

AI tools help early dementia detection

Scientists from the Universities of Edinburgh and Dundee are?launching?a massive AI-driven study of over 1.6 million brain scans to develop tools for early dementia prediction and diagnosis.

• The project, called NEURii, will use AI and machine learning to analyze CT and MRI scans from Scottish patients over the past decade.
• Researchers aim to create digital tools for radiologists to assess dementia risk during routine scans.
• The study will match image data with linked health records to identify patterns associated with dementia risk.

We've seen AI developing new cancer drugs,?3D printing?lifelike human organs, and now creating tools for early dementia detection. As AI rapidly advances in healthcare, we're accelerating into a new era of personalized medicine and preventative care.

AI spots cancer earlier than ever

Researchers recently?developed?an AI tool called AINU that can differentiate cancer cells from normal cells and detect early stages of viral infection, by analyzing high-resolution images of cell nuclei.

• AINU uses a convolutional neural network to analyze images captured by STORM microscopy, which offers nanoscale resolution.
• The AI can detect structural changes in cells as small as 20 nanometers, 5,000 times smaller than a human hair's width.
• AINU also detected viral infections (herpes simplex virus type-1) just one hour after infection by observing subtle changes in DNA packing.
• The tool can accurately identify stem cells too, which could accelerate stem cell research without relying on animal testing.

Clinical applications may be years away, but AI healthcare breakthroughs like AINU are only accelerating — and will dramatically revolutionize scientific research in the coming years.

?? Reflections and Insights ??

Is AI in a bubble right now?

A comprehensive analysis by Kelvin Mu at Translink Capital comparing the current AI cycle to the internet/telecom cycle of the 90s. He examines the technological, economic, and capital differences between the two eras and concludes that while a bubble may be inevitable in the long run, we are still far from reaching that point.

America & China's Chip Race

The U.S. is pursuing aggressive measures to strengthen domestic semiconductor production through the CHIPS Act and sanctions aimed at crippling China's tech advancement. Despite these efforts, China's semiconductor buildout is surging with near-record imports of manufacturing equipment and growing domestic chip production. This intensifying race suggests a continuous geopolitical tug-of-war over semiconductor supply chain dominance.

AI Is a Services Revolution

This article discusses the impact of LLMs on the service economy, particularly on knowledge work sectors such as legal, healthcare, and education. While rapid AI advancements suggest significant automation potential, the adoption by enterprises is slow, with many still in the experimentation phase. Actual deployment is expected to be gradual. Specialized AI startups that leverage LLMs to improve industry-specific workflows will have a competitive advantage in the evolving market.

Tech Giants Drive AI Revolution: Transforming Business Models and User Experiences

Major technology companies are increasingly integrating artificial intelligence (AI) into their core products and services, creating new revenue streams and enhancing user experiences. This shift is transforming business models and consumer interactions across various sectors, with AI becoming a key differentiator. Companies like Google, Microsoft, and Amazon are leading this trend, embedding AI into search engines, productivity tools, and personalized recommendations.

Industry experts highlight both the potential and challenges of AI adoption. AI can automate processes, improve customer insights, and streamline operations, offering a competitive edge. However, financial barriers such as high API fees, the need for frequent hardware upgrades, and data management challenges pose significant hurdles, particularly for smaller businesses. Effective AI deployment requires well-prepared, actionable data, making data preparation a critical step for businesses looking to leverage AI.

?? Stay Updated: Receive regular updates delivered straight to your inbox, ensuring you're always in the loop with the latest AI developments. Don't miss out on the opportunity to be at the forefront of innovation!

?? Ready to Unleash the Power of AI? Subscribe Now and Let the Insights Begin! ??