Pocket devices with AI inside - Weekly News Roundup - Issue #450
Plus: DeepMind's new Olympiad-level AI system for geometry; Tesla Bot folds a shirt; OpenAI quietly drops a ban on military usage; Meta promises open-source AGI; and more!
Welcome to Weekly News Roundup Issue #450. This week, we will take a closer look at AI-powered devices shown at CES 2024 and what they can tell us about the future of consumer electronics.
In other news, DeepMind has released a new AI system capable of Olympiad-level geometry, Meta has announced plans to open-source AGI, and OpenAI now expresses openness to collaborations with the military. Additionally, Tesla's Bot demonstrated its capability to fold a shirt, and humanoid robots are being introduced to a BMW factory. Meanwhile, the FDA has approved a new CRISPR gene-editing treatment, and a brain-dead man was connected to a gene-edited pig liver for three days.
I hope you enjoy Issue #450 of the Weekly News Roundup!
Last year, as AI burst onto the scene, many people started to experiment with how this new, exciting generative AI technology can be used to create new products and services. And when most people were developing software, some asked themselves what kind of new hardware devices could be built now. At CES 2024, we got a glimpse into the future with pocket-sized personal assistants that are as simple to use as talking to a human.
The dream of being able to just talk to devices is not new. In the last decade, we have seen attempts to make the voice interface as seamless as possible with voice assistants such as Siri or Alexa and various smart speakers. However, these efforts were not good enough and failed. Apple’s Homepod wasn’t a successful product, Google Assistant and Siri are still away from what they were promised to be and Amazon has reportedly lost over $10 billion developing Alexa and Echo. But the introduction of large language models and how good they are with natural language reignited the dream of creating personal assistants that you can just speak with. And there is no place better than CES to show pioneering devices.
The star of the CES 2024 was Rabbit R1, an orange device with a built-in AI assistant. The idea behind R1 is to create the simplest possible computer that does not have any apps and is operated only with voice - you just say what you want the device with an onboard AI will do the thing. R1 received a lot of attention and sold two pre-order batches (20,000 devices in total, each costing $199) in just two days. At the time I am writing this, R1 is on its fifth pre-order batch.
Rabbit R1 is not the only or even the first device that is going full-on into this new class of devices operated with natural language. The first one to gather significant public attention was Humane’s AI Pin. This device, starting at $699 and released late last year, doesn’t even have a traditional screen. Instead, it relies on an interface combining voice control, a camera, gestures, and a small built-in projector. AI Pin, however, does not have a built-in AI model - it uses OpenAI models. Other AI-powered devices shown at CES 2024 included TranscribeGlass (smart glasses transcribing and displaying the conversation on a small screen ala Google Glass) or TimeKettel’s translation devices.
It is not just small startups exploring how to integrate language models and hardware - tech giants are also working on it, too. There were reports last year that Jony Ive, the legendary industrial designer behind some of Apple’s biggest products, has met with OpenAI’s CEO Sam Altman to discuss “the iPhone for artificial intelligence.” Nothing more is known beyond that and there were no new rumours or leaks about OpenAI’s hypothetical first consumer device. Meanwhile, Apple and Google are focusing on improving their own voice assistants rather than entering a new category of devices. Apple is apparently working on its large language model codenamed Ajax, which is currently being tested internally by Apple employees. Rumours are that Ajax will be used to greatly improve Siri’s performance. Apple’s new AI products are expected to be revealed at WWDC this year. Google is doing the same by including Gemini Nano (the smallest model from the Gemini family of language models) in Pixel 8 Pro phones to enhance Google Assistant’s performance.
Both Apple and Google will run their large language models directly on the phone. Google does that already with Gemini Nano and we can expect Apple will do the same. Running language models on a small device might bring with it changes to how chips powering these devices are designed. Over the past few years, we have seen an increasing number of specialized accelerators being incorporated into chips, including dedicated AI accelerators. Apple has been doing that for years now with Neural Engine on all of their devices and others are following suit. Qualcomm is including dedicated AI silicon in their chips and AMD is doing the same for PCs. The need for even more efficient AI chips may require the adoption of new architectures, like in-memory computing, neuromorphic computing or perhaps something completely new.
Rabbit R1 or Humane AI Pin are not guaranteed to succeed. I see them as interesting ideas, prototypes bought by very early adopters. There is a big chance they and others like them will fail or at most become a niche product. However, what these projects do is to show us the possible path the consumer technology can take. And the path seems to be adding more and more intelligence to our devices. Maybe this time they will deserve to be called “smart”.
If you enjoy this post, please click the ❤️ button or share it.
I warmly welcome all new subscribers to the newsletter this week. I’m happy to have you here and I hope you’ll enjoy my work. A heartfelt thank you goes to everyone who joined as paid subscribers this week.
The best way to support the Humanity Redefined newsletter is by becoming a paid subscriber.
If you enjoy and find value in my writing, please hit the like button and share your thoughts in the comments. Additionally, please consider sharing this newsletter with others who might also find it valuable.
For those who prefer to make a one-off donation, you can 'buy me a coffee' via Ko-fi. Every coffee bought is a generous support towards the work put into this newsletter.
Your support, in any form, is deeply appreciated and goes a long way in keeping this newsletter alive and thriving.
🦾 More than a human
▶️ 'A vaccine for aging in under ten years' (14:03)
Longevity.Technology recently interviewed Ronjon Nag, founder of Agemica, which is developing a vaccine for ageing to make a dramatic difference to lifespan and healthspan. The company hopes its omics-based tech will extend both lifespan and healthspan, and the platform has already been validated in 17 cancer areas and predicted antiaging candidates that have higher efficacy, reduced side effects and lower chances of developing drug resistance. The big goal is to have this vaccine within the next 10 years, which Nag believes is possible with machine learning.
A brain-dead man was attached to a gene-edited pig liver for three days
A team of researchers successfully conducted an experiment in which a genetically engineered pig liver was used to assist a brain-dead human. This external liver, connected to the man's body through a machine, functioned for three days. This achievement marks a significant advancement in treating acute liver failure, which can result from infection, poisoning, or excessive alcohol consumption. The experiment also demonstrates the potential for short-term use of animal organs acting more like a piece of machinery than a traditional transplant.
🔮 Future visions
▶️ Monika Bielskyte - Protopia vs TESCREAL Visions of Tech Rapture (57:45)
I really enjoyed this talk by Monika Bielskyte which asks us which future we want - the dystopian, cyberpunk future or a more hopeful, inclusive future which Monika calls Protopia but I know better as solarpunk. The talk is an interesting critique of the mainstream futurism known as TESCREAL (Transhumanism, Extropianism, Singularitarianism, Cosmism, Rationalism, Effective Altruism, and Longtermism) and presents a different vision of the future based on plurality, interdependence, embodiment, biosphere regeneration, creative flourishment, spiritual fulfilment, and evolved values.
🧠 Artificial Intelligence
AlphaGeometry: An Olympiad-level AI system for geometry
Researchers from DeepMind have developed AlphaGeometry, a new AI capable of solving geometric problems at the level of top human International Mathematical Olympiad gold medalists. In a benchmark test featuring 30 Olympiad geometry problems, AlphaGeometry successfully solved 25 within the standard Olympiad time limit. For comparison, the previous state-of-the-art system solved only 10 of these problems, while the average human gold medalist solved approximately 25.9 problems. AlphaGeometry is a neuro-symbolic system, combining a language model with a symbolic deduction engine. These components work in tandem to find proofs for complex geometry theorems. The system's architecture embodies the concept of 'thinking, fast and slow', with one part providing quick, 'intuitive' ideas and the other offering more deliberate, rational decision-making. This approach is indicative of the direction in which AI research is heading towards artificial general intelligence. For more information, the AlphaGeometry paper is available in Nature, and the code can be found on GitHub.
OpenAI Quietly Deletes Ban on Using ChatGPT for “Military and Warfare”
OpenAI has discreetly altered its usage policy, opening the door for its products to potentially be used by military organizations. Until January 10, OpenAI's 'usage policies' page explicitly prohibited 'activities that have a high risk of physical harm,' including specifically 'weapons development' and 'military and warfare,' as reported by The Intercept, which first broke the story. While OpenAI still does not allow the use of its AI models to cause harm to people, these recent changes now categorize military organizations as potential customers. Bloomberg reports that the US military has projects, such as cybersecurity and tools to assist veterans, for which it wants to use OpenAI tools. This shift could be the first step towards OpenAI becoming a part of the military-tech complex.
Mark Zuckerberg’s new goal is creating artificial general intelligence
Mark Zuckerberg has announced on Instagram Meta's plans to develop artificial general intelligence (AGI) and then open-source it. He also revealed plans to acquire 340,000 of Nvidia’s top-of-the-line H100 GPUs. This acquisition will increase the total number of GPUs available to Meta’s researchers to nearly 600,000 by the end of 2024. In addition, Zuckerberg mentioned that Meta is currently training Llama 3, the successor to one of the best and most popular open-source large language models. “Llama 2 wasn’t an industry-leading model, but it was the best open-source model,” he says. “With Llama 3 and beyond, our ambition is to build things that are at the state of the art and eventually the leading models in the industry.”
Jeffrey Katzenberg Says A.I. Will Eliminate 90 Percent of Artist Jobs on Animated Films
Jeffrey Katzenberg, co-founder of DreamWorks SKG, predicts that AI will replace 90% of animation artists, leading to drastic changes in the industry. Speaking at the Bloomberg New Economy Forum, he highlighted AI's impact, calling to think of AI technology as a creative tool, akin to a new paintbrush or a new camera. Katzenberg believes that while AI will significantly reduce the workforce required for animation projects, individual creativity will remain crucial for directing AI.
AI will affect 40% of jobs and probably worsen inequality, says IMF head
A new report from the International Monetary Fund (IMF) warns that AI could affect 40% of global jobs, with advanced economies being at higher risk. About 60% of jobs in these economies are exposed to AI, potentially impacting half of them negatively. However, AI may also enhance productivity in jobs that work with a high degree of human interaction, like surgery and law, which have a high "complementarity" with AI. The IMF stresses the importance of social safety nets and retraining programs to manage the transition and address potential inequality increases. High-wage earners might benefit more from AI, further amplifying income disparities.
If you're enjoying the insights and perspectives shared in the Humanity Redefined newsletter, why not spread the word?
🤖 Robotics
▶️ Optimus folds a shirt (0:31)
Elon Musk shared a video of Tesla’s humanoid robot, Optimus, folding a shirt. Musk later clarified that the robot had not yet achieved this autonomously. Nevertheless, the video showcases the dexterity and capabilities of the robot. Expect to see more of this kind of showcases as the commercial humanoid robots are getting closer to being widely available.
BMW will deploy Figure’s humanoid robot at South Carolina plant
Figure, one of the companies promising to introduce commercial humanoid robots soon, announced a “commercial agreement” with BMW to bring its first humanoid robot, the Figure 01, to a BMW manufacturing facility in South Carolina. As reported by TechCrunch, the number of Figure 01 robots to be initially deployed was not disclosed by either Figure or BMW. It is known that the robot will initially do five tasks, which will be introduced one at a time.
IFR: World sets record for operational robots in 2022
The Robot Report reveals that a record 3.9 million robots were operational in 2022, according to the World Robotics report by the International Federation of Robotics (IFR). This report highlights the global robot density (a measure of industrial robots per 10,000 employees in a country) reached a new peak of 151 robots. Leading the list as the most automated country was South Korea, with 1,012 robots per 10,000 employees, followed by Singapore, Germany, Japan, and China.
NASA’s robotic, self-assembling structures could be the next phase of space construction
NASA presents ARMADAS (Automated Reconfigurable Mission Adaptive Digital Assembly Systems), a self-assembling robotic structure, a potential solution for off-planet housing. The system consists of two types of robots: one that carries building blocks (voxels) and another that fastens them together. These robots, which can work continuously and don't require high precision, have successfully assembled a shelter structure from 256 voxels over 4.2 days. Future enhancements include faster and more reliable robots, and the system might eventually use in-situ materials. The research aims to explore larger structures using swarms of robots, envisioning autonomous recharging and possibly wireless power beaming to support their operations.
🧬 Biotechnology
FDA Approves New CRISPR Gene-Editing Treatment
The U.S. Food and Drug Administration recently approved Casgevy, a therapy developed by Vertex Pharmaceuticals and CRISPR Therapeutics, for treating transfusion-dependent beta-thalassemia using CRISPR gene-editing technology. This approval marks the second major U.S. regulatory nod for CRISPR technology, following its approval for treating sickle cell disease. Vertex Pharmaceuticals is working to establish a network of treatment centres across the U.S. for administering Casgevy, with nine centres currently active and more expected soon.
Cloned rhesus monkey lives to adulthood for first time
For the first time, a cloned rhesus monkey has survived into adulthood, living for over two years and marking a significant milestone in cloning research. This result was achieved using a novel approach that differs from traditional cloning techniques like those used for Dolly the sheep and opens new possibilities for using cloned primates in drug testing and behavioural research. However, the cloning process, particularly in primates, still faces challenges with low birth and survival rates, as well as inefficiencies in the cloning technique.
Scientists Extend Life Span in Mice by Restoring This Brain-Body Connection
A new study has revealed that a specific group of neurons in the hypothalamus, marked by the protein Ppp1r17, plays a crucial role in longevity by maintaining a communication link with fatty tissues. In young mice, these neurons regulate fat cells to release energy, but this connection deteriorates with age, leading to health and mobility decline. By genetically altering aged mice to reactivate this neuron-fat communication, researchers observed an extended lifespan by about seven percent and improved health. This discovery not only expands our understanding of brain-body interactions in ageing, but also suggests promising new ways for further research into optimizing the fat-to-brain feedback loop to enhance healthy ageing.
Thanks for reading. If you enjoyed this post, please click the ❤️ button or share it.
Humanity Redefined sheds light on the bleeding edge of technology and how advancements in AI, robotics, and biotech can usher in abundance, expand humanity's horizons, and redefine what it means to be human.
A big thank you to my paid subscribers, to my Patrons: whmr, Florian, dux, Eric, Preppikoma and Andrew, and to everyone who supports my work on Ko-Fi. Thank you for the support!