A Paralyzed Man Can Walk Naturally Again With Brain and Spine Implants
It is a “digital bridge” between the brain and the spinal cord, bypassing injured sections. They use a recursive exponentially weighted Markov-switching multilinear model algorithm to decode the intention to perform lower limb movements.
New York Times: https://archive.is/PMnEi
Nature: https://archive.is/98uMj
Paper: https://www.nature.com/articles/s41586-023-06094-5
It is a “digital bridge” between the brain and the spinal cord, bypassing injured sections. They use a recursive exponentially weighted Markov-switching multilinear model algorithm to decode the intention to perform lower limb movements.
New York Times: https://archive.is/PMnEi
Nature: https://archive.is/98uMj
Paper: https://www.nature.com/articles/s41586-023-06094-5
❤10
How often fascism was mentioned in Pravda: https://twitter.com/whyvert/status/1661526846201032706
👍5😁1
Videos showing Ukrainian maritime drones attacking the Ivan Khurs SIGINT intelligence collection ship 400 km from the Ukrainian shore.
🔥8😁1😱1
Links for 2023-05-26
1. New superbug-killing antibiotic discovered using AI https://news.mit.edu/2023/using-ai-scientists-combat-drug-resistant-infections-0525
2. Improving Factuality and Reasoning in Language Models through Multiagent Debate https://composable-models.github.io/llm_debate/
3. Repurposing LLMs As Both World Model & Reasoning Agent: “On plan generation, numerical reasoning, and logical reasoning tasks, RAP (our framework) outperforms previous methods including CoT/Least-to-Most + self-consistency. Specifically, RAP (LLaMA-33B) even outperforms GPT-4 in a setting of Blocksworld.” https://arxiv.org/abs/2305.14992
4. An efficient finetuning approach that reduces memory usage enough to finetune a 65B model on a single 48GB GPU while preserving full 16-bit finetuning task performance. https://arxiv.org/abs/2305.14314
5. Goat: Fine-tuned LLaMA Outperforms GPT-4 on Arithmetic Tasks https://arxiv.org/abs/2305.14201
6. Resolving code review comments with ML: “…we describe applying recent advances of large sequence models in a real-world setting to automatically resolve code review comments in the day-to-day development workflow at Google…” https://ai.googleblog.com/2023/05/resolving-code-review-comments-with-ml.html?m=1
7. “GPT-4/PaLM-2 have both shown almost perfect performance on existing grade school math dataset. What about more challenging STEM questions, especially the ones which require specific theorems, like Stoke's theorem, Wiener Process, etc? ... We found that GPT-4 is performing surprisingly well on TheoremQA. It achieves 51% accuracy with PoT and 43% with CoT. All the existing open-source LLMs are only getting 10-14% accuracy.” https://twitter.com/WenhuChen/status/1660832837715611648
8. “Introducing LeTI, a new LM finetuning paradigm that explores LMs' potential to learn from textual interactions & feedback, allowing LMs to understand not just if they were wrong, but why.” https://twitter.com/xingyaow_/status/1659246169950650391
9. Can we use LLM to Speedup LLM inference? https://zhengzangw.github.io/blogs/seqsch/
10. A Generalist Dynamics Model for Control https://arxiv.org/abs/2305.10912
11. Apple’s upcoming Personal Voice feature can clone your voice in 15 minutes using AI https://www.macworld.com/article/1919264/ios-17-macos-14-accessibility-live-speech-personal-voice.html
12. "While all attention was focused on Neuralink, Paradromics quietly developed a breakthrough device and technology" https://www.prnewswire.com/news-releases/paradromics-raises-33-million-in-funding-achieves-breakthrough-medical-device-designation-from-fda-301827969.html
13. Once again, ideology distorts science: The editor-in-chief of Scientific American flubs big time, wrongly asserting that sparrows have four sexes. https://whyevolutionistrue.com/2023/05/18/once-again-ideology-distorts-science-the-editor-in-chief-of-scientific-american-flubs-big-time-wrongly-asserting-that-sparrows-have-four-sexes/
14. The Oldest Known Blueprints Depict Stone Age ‘Megastructures’ https://www.vice.com/en/article/pkagbb/the-oldest-known-blueprints-depict-stone-age-megastructures
1. New superbug-killing antibiotic discovered using AI https://news.mit.edu/2023/using-ai-scientists-combat-drug-resistant-infections-0525
2. Improving Factuality and Reasoning in Language Models through Multiagent Debate https://composable-models.github.io/llm_debate/
3. Repurposing LLMs As Both World Model & Reasoning Agent: “On plan generation, numerical reasoning, and logical reasoning tasks, RAP (our framework) outperforms previous methods including CoT/Least-to-Most + self-consistency. Specifically, RAP (LLaMA-33B) even outperforms GPT-4 in a setting of Blocksworld.” https://arxiv.org/abs/2305.14992
4. An efficient finetuning approach that reduces memory usage enough to finetune a 65B model on a single 48GB GPU while preserving full 16-bit finetuning task performance. https://arxiv.org/abs/2305.14314
5. Goat: Fine-tuned LLaMA Outperforms GPT-4 on Arithmetic Tasks https://arxiv.org/abs/2305.14201
6. Resolving code review comments with ML: “…we describe applying recent advances of large sequence models in a real-world setting to automatically resolve code review comments in the day-to-day development workflow at Google…” https://ai.googleblog.com/2023/05/resolving-code-review-comments-with-ml.html?m=1
7. “GPT-4/PaLM-2 have both shown almost perfect performance on existing grade school math dataset. What about more challenging STEM questions, especially the ones which require specific theorems, like Stoke's theorem, Wiener Process, etc? ... We found that GPT-4 is performing surprisingly well on TheoremQA. It achieves 51% accuracy with PoT and 43% with CoT. All the existing open-source LLMs are only getting 10-14% accuracy.” https://twitter.com/WenhuChen/status/1660832837715611648
8. “Introducing LeTI, a new LM finetuning paradigm that explores LMs' potential to learn from textual interactions & feedback, allowing LMs to understand not just if they were wrong, but why.” https://twitter.com/xingyaow_/status/1659246169950650391
9. Can we use LLM to Speedup LLM inference? https://zhengzangw.github.io/blogs/seqsch/
10. A Generalist Dynamics Model for Control https://arxiv.org/abs/2305.10912
11. Apple’s upcoming Personal Voice feature can clone your voice in 15 minutes using AI https://www.macworld.com/article/1919264/ios-17-macos-14-accessibility-live-speech-personal-voice.html
12. "While all attention was focused on Neuralink, Paradromics quietly developed a breakthrough device and technology" https://www.prnewswire.com/news-releases/paradromics-raises-33-million-in-funding-achieves-breakthrough-medical-device-designation-from-fda-301827969.html
13. Once again, ideology distorts science: The editor-in-chief of Scientific American flubs big time, wrongly asserting that sparrows have four sexes. https://whyevolutionistrue.com/2023/05/18/once-again-ideology-distorts-science-the-editor-in-chief-of-scientific-american-flubs-big-time-wrongly-asserting-that-sparrows-have-four-sexes/
14. The Oldest Known Blueprints Depict Stone Age ‘Megastructures’ https://www.vice.com/en/article/pkagbb/the-oldest-known-blueprints-depict-stone-age-megastructures
👍3
Voyager: An Open-Ended Embodied Agent with Large Language Models https://voyager.minedojo.org/
- The first LLM-powered embodied lifelong learning agent in Minecraft that continuously explores the world, acquires diverse skills, and makes novel discoveries without human intervention.
- Empirically, Voyager shows strong in-context lifelong learning capability and exhibits exceptional proficiency in playing Minecraft. It obtains 3.3x more unique items, travels 2.3x longer distances, and unlocks key tech tree milestones up to 15.3x faster than prior SOTA.
- Voyager is able to utilize the learned skill library in a new Minecraft world to solve novel tasks from scratch, while other techniques struggle to generalize.
- The first LLM-powered embodied lifelong learning agent in Minecraft that continuously explores the world, acquires diverse skills, and makes novel discoveries without human intervention.
- Empirically, Voyager shows strong in-context lifelong learning capability and exhibits exceptional proficiency in playing Minecraft. It obtains 3.3x more unique items, travels 2.3x longer distances, and unlocks key tech tree milestones up to 15.3x faster than prior SOTA.
- Voyager is able to utilize the learned skill library in a new Minecraft world to solve novel tasks from scratch, while other techniques struggle to generalize.
❤5🔥3
Links for 2023-05-27
1. The market is upping its evaluation of AI econ impact: “Nvidia Shares Jump as Chip Maker Approaches $1 Trillion Valuation” [Financial Times] https://archive.is/XLOKN
2. An innovative open-source conversational LLM engineered specifically for healthcare https://arxiv.org/abs/2305.12031
3. Capabilities of GPT-4 on Medical Challenge Problems https://arxiv.org/abs/2303.13375
4. Large Language Models are Few-Shot Health Learners https://arxiv.org/abs/2305.15525
5. Dr. GPT Will See You Now. How will AI change medicine? https://www.city-journal.org/article/how-will-ai-change-medicine
6. “These new results follow a recent trend of places where neuroscience results are starting to have parallels with results found in neural networks!” https://twitter.com/ch402/status/1659284365380313088
7. CodeCompose: A Large-Scale Industrial Deployment of AI-assisted Code Authoring https://arxiv.org/abs/2305.12050
8. “We introduce a method that combines a Transformer mutation model with Monte-Carlo Tree Search.” https://arxiv.org/abs/2302.11223
9. Recent advances in Hopfield networks of associative memory may be the guiding theoretical principle for designing novel large scale neural architectures. https://www.nature.com/articles/s42254-023-00595-y
10. Take Your Ultrawide Monitors Everywhere With an AR Laptop [Wired] https://archive.is/sOByP
11. Novel 3D printing method a ‘game changer’ for discovery, manufacturing of new materials https://news.nd.edu/news/novel-3d-printing-method-a-game-changer-for-discovery-manufacturing-of-new-materials/
12. Team uses 3D printing to strengthen a key material in aerospace, energy-generation applications https://news.mit.edu/2023/team-uses-3d-printing-strengthen-material-key-applications-aerospace-energy-generation-0519
13. Self-driving laboratories to autonomously navigate the protein fitness landscape https://www.biorxiv.org/content/10.1101/2023.05.20.541582v1
14. The man who discovered the first 70 of the 100 known exoplanets (planets that orbit stars other than the sun) has not only been 'cancelled,' his name is now being scrubbed from papers & his colleagues are being targeted for cancellation as well. https://quillette.com/2023/05/25/campus-puritans/
1. The market is upping its evaluation of AI econ impact: “Nvidia Shares Jump as Chip Maker Approaches $1 Trillion Valuation” [Financial Times] https://archive.is/XLOKN
2. An innovative open-source conversational LLM engineered specifically for healthcare https://arxiv.org/abs/2305.12031
3. Capabilities of GPT-4 on Medical Challenge Problems https://arxiv.org/abs/2303.13375
4. Large Language Models are Few-Shot Health Learners https://arxiv.org/abs/2305.15525
5. Dr. GPT Will See You Now. How will AI change medicine? https://www.city-journal.org/article/how-will-ai-change-medicine
6. “These new results follow a recent trend of places where neuroscience results are starting to have parallels with results found in neural networks!” https://twitter.com/ch402/status/1659284365380313088
7. CodeCompose: A Large-Scale Industrial Deployment of AI-assisted Code Authoring https://arxiv.org/abs/2305.12050
8. “We introduce a method that combines a Transformer mutation model with Monte-Carlo Tree Search.” https://arxiv.org/abs/2302.11223
9. Recent advances in Hopfield networks of associative memory may be the guiding theoretical principle for designing novel large scale neural architectures. https://www.nature.com/articles/s42254-023-00595-y
10. Take Your Ultrawide Monitors Everywhere With an AR Laptop [Wired] https://archive.is/sOByP
11. Novel 3D printing method a ‘game changer’ for discovery, manufacturing of new materials https://news.nd.edu/news/novel-3d-printing-method-a-game-changer-for-discovery-manufacturing-of-new-materials/
12. Team uses 3D printing to strengthen a key material in aerospace, energy-generation applications https://news.mit.edu/2023/team-uses-3d-printing-strengthen-material-key-applications-aerospace-energy-generation-0519
13. Self-driving laboratories to autonomously navigate the protein fitness landscape https://www.biorxiv.org/content/10.1101/2023.05.20.541582v1
14. The man who discovered the first 70 of the 100 known exoplanets (planets that orbit stars other than the sun) has not only been 'cancelled,' his name is now being scrubbed from papers & his colleagues are being targeted for cancellation as well. https://quillette.com/2023/05/25/campus-puritans/
Links for 2023-05-28
1. Impressive: GPT-4 Out-performs RL Algorithms by Studying Papers and Reasoning https://arxiv.org/abs/2305.15486 (apparently this was already known in 2012(!!!) but forgotten/ignored: http://people.csail.mit.edu/branavan/papers/acl2012.pdf)
2. We are on the cusp of a remarkable transformation in the way humans explore outer space—potentially the biggest change in spaceflight since the Soviet Union launched the Sputnik satellite in 1957. https://arstechnica.com/space/2023/05/at-long-last-the-glorious-future-we-were-promised-in-space-is-on-the-way/
3. “ImageBind brings machines one step closer to the human ability to bind together information from many different senses. It's the first AI model capable of binding data from six modalities at once.” https://ai.facebook.com/blog/imagebind-six-modalities-binding-ai/
4. AI is moving so fast, fintech startups are having to quickly pivot. [New York Times] https://archive.is/hBywW
5. Overview of LangChain — Motto: Building applications with LLMs through composability https://medium.com/technology-hits/overview-of-langchain-9f6362707cd0
6. RWKV: Reinventing RNNs for the Transformer Era https://arxiv.org/abs/2305.13048
7. Automatic Prompt Optimization with "Gradient Descent" and Beam Search https://arxiv.org/abs/2305.03495
8. Bayes is guaranteed to overfit, for any model, any prior, and every data point https://www.yulingyao.com/blog/2023/overfit/
9. Metalenses are finally moving into consumers’ hands — Future metalens applications will take advantage of the technology’s ability to detect polarized light. https://spectrum.ieee.org/metalens-2660294513
10. Noise is a secret destroyer of productivity. https://twitter.com/emollick/status/1660093233181933568
11. Standardized exams measure intrinsic ability, not racial or socioeconomic privilege https://milkyeggs.com/society/standardized-exams-measure-intrinsic-ability-not-racial-or-socioeconomic-privilege/
12. “…a large and robust negative relationship between length of Muslim rule and literacy” https://marginalrevolution.com/marginalrevolution/2023/05/islam-and-human-capital-in-historical-spain.html
1. Impressive: GPT-4 Out-performs RL Algorithms by Studying Papers and Reasoning https://arxiv.org/abs/2305.15486 (apparently this was already known in 2012(!!!) but forgotten/ignored: http://people.csail.mit.edu/branavan/papers/acl2012.pdf)
2. We are on the cusp of a remarkable transformation in the way humans explore outer space—potentially the biggest change in spaceflight since the Soviet Union launched the Sputnik satellite in 1957. https://arstechnica.com/space/2023/05/at-long-last-the-glorious-future-we-were-promised-in-space-is-on-the-way/
3. “ImageBind brings machines one step closer to the human ability to bind together information from many different senses. It's the first AI model capable of binding data from six modalities at once.” https://ai.facebook.com/blog/imagebind-six-modalities-binding-ai/
4. AI is moving so fast, fintech startups are having to quickly pivot. [New York Times] https://archive.is/hBywW
5. Overview of LangChain — Motto: Building applications with LLMs through composability https://medium.com/technology-hits/overview-of-langchain-9f6362707cd0
6. RWKV: Reinventing RNNs for the Transformer Era https://arxiv.org/abs/2305.13048
7. Automatic Prompt Optimization with "Gradient Descent" and Beam Search https://arxiv.org/abs/2305.03495
8. Bayes is guaranteed to overfit, for any model, any prior, and every data point https://www.yulingyao.com/blog/2023/overfit/
9. Metalenses are finally moving into consumers’ hands — Future metalens applications will take advantage of the technology’s ability to detect polarized light. https://spectrum.ieee.org/metalens-2660294513
10. Noise is a secret destroyer of productivity. https://twitter.com/emollick/status/1660093233181933568
11. Standardized exams measure intrinsic ability, not racial or socioeconomic privilege https://milkyeggs.com/society/standardized-exams-measure-intrinsic-ability-not-racial-or-socioeconomic-privilege/
12. “…a large and robust negative relationship between length of Muslim rule and literacy” https://marginalrevolution.com/marginalrevolution/2023/05/islam-and-human-capital-in-historical-spain.html
👍8⚡1
Links for 2023-05-29
1. Someone tried out all 134 ChatGPT plugins https://twitter.com/aakashg0/status/1662313397776691200
2. How AI can help discover algorithms https://www.youtube.com/watch?v=fDAPJ7rvcUw
3. Interview with Jürgen Schmidhuber, ‘Father Of Modern AI’, who says his life’s work won’t lead to dystopia. https://www.reddit.com/r/MachineLearning/comments/13q6k4a/interview_with_juergen_schmidhuber_renowned/
4. Investors are so horny for AI that they’ve invested $175mn in OpenAI’s VC fund [Financial Times] https://archive.is/TOd5p
5. Corporate VCs ride AI startup wave https://www.axios.com/2023/05/27/corporate-vcs-ride-ai-startup-wave
6. Money pours into AI startups that save countless hours of legal paperwork https://www.semafor.com/article/05/24/2023/investors-back-ai-startups-that-save-hours-of-legal-paperwork
7. Why are so many giants of AI getting GPTs so badly wrong? https://medium.com/@fergal.reid/why-are-so-many-giants-of-ai-getting-gpts-so-badly-wrong-f8dadfac4f61
8. “If you told me that the aliens had a single memory system for brains and genes, that they encoded skills in their version of DNA and passed it on heritably or even horizontally by sharing blood, my expectations for their human-style niceness would drop, though not to zero. Those sound like aliens that might end up with a different empathy-architecture that didn't do sympathy.” https://twitter.com/ESYudkowsky/status/1660623336567889920
9. A stylized dialogue on John Wentworth's claims about markets and optimization https://www.lesswrong.com/posts/fJBTRa7m7KnCDdzG5/a-stylized-dialogue-on-john-wentworth-s-claims-about-markets
10. Rectangular log map-scheme of the Observable Universe https://www.pablocarlosbudassi.com/2021/02/atlas-of-universe-is-linear-version-of_15.html
11. Circassian genocide https://en.m.wikipedia.org/wiki/Circassian_genocide
1. Someone tried out all 134 ChatGPT plugins https://twitter.com/aakashg0/status/1662313397776691200
2. How AI can help discover algorithms https://www.youtube.com/watch?v=fDAPJ7rvcUw
3. Interview with Jürgen Schmidhuber, ‘Father Of Modern AI’, who says his life’s work won’t lead to dystopia. https://www.reddit.com/r/MachineLearning/comments/13q6k4a/interview_with_juergen_schmidhuber_renowned/
4. Investors are so horny for AI that they’ve invested $175mn in OpenAI’s VC fund [Financial Times] https://archive.is/TOd5p
5. Corporate VCs ride AI startup wave https://www.axios.com/2023/05/27/corporate-vcs-ride-ai-startup-wave
6. Money pours into AI startups that save countless hours of legal paperwork https://www.semafor.com/article/05/24/2023/investors-back-ai-startups-that-save-hours-of-legal-paperwork
7. Why are so many giants of AI getting GPTs so badly wrong? https://medium.com/@fergal.reid/why-are-so-many-giants-of-ai-getting-gpts-so-badly-wrong-f8dadfac4f61
8. “If you told me that the aliens had a single memory system for brains and genes, that they encoded skills in their version of DNA and passed it on heritably or even horizontally by sharing blood, my expectations for their human-style niceness would drop, though not to zero. Those sound like aliens that might end up with a different empathy-architecture that didn't do sympathy.” https://twitter.com/ESYudkowsky/status/1660623336567889920
9. A stylized dialogue on John Wentworth's claims about markets and optimization https://www.lesswrong.com/posts/fJBTRa7m7KnCDdzG5/a-stylized-dialogue-on-john-wentworth-s-claims-about-markets
10. Rectangular log map-scheme of the Observable Universe https://www.pablocarlosbudassi.com/2021/02/atlas-of-universe-is-linear-version-of_15.html
11. Circassian genocide https://en.m.wikipedia.org/wiki/Circassian_genocide
IQ scores by ethnic group in a nationally-representative sample of 10-year old American children https://humanvarieties.org/2023/05/27/iq-scores-by-ethnic-group-in-a-nationally-representative-sample-of-10-year-old-american-children/
🔥9👎4👍2⚡1
This media is not supported in your browser
VIEW IN TELEGRAM
I was initially skeptical that this video was real, but the following videos suggest that it may very well be:
1. https://www.youtube.com/watch?v=cQYZgNlrLA8
2. https://www.youtube.com/shorts/uL5rlI19MFQ
3. https://youtu.be/mILCMQJkOBA
4. https://youtu.be/o8kG_rgPbGA
5. https://youtu.be/Fzxj1rnCUUg
1. https://www.youtube.com/watch?v=cQYZgNlrLA8
2. https://www.youtube.com/shorts/uL5rlI19MFQ
3. https://youtu.be/mILCMQJkOBA
4. https://youtu.be/o8kG_rgPbGA
5. https://youtu.be/Fzxj1rnCUUg
🤯13👍1
Links for 2023-05-30
1. Gemini will bring the next big timeline update https://www.lesswrong.com/posts/zhf7LrMmnFbJKqqPL/gemini-will-bring-the-next-big-timeline-update
2. Training LLMs via Simulated Human Societies https://arxiv.org/abs/2305.16960
3. Mindstorms in Natural Language-Based Societies of Mind https://arxiv.org/abs/2305.17066
4. Think Before You Act: Decision Transformers with Internal Working Memory https://arxiv.org/abs/2305.16338
5. LayoutGPT: Outperforms text-to-image models/systems by 20-40% and achieves comparable performance as human users in designing visual layouts for numerical and spatial correctness. https://arxiv.org/abs/2305.15393
6. Under 1,350 square miles of jungle in Guatemala, scientists discovered 417 cities dating back to circa 1,000 B.C. and that are connected by nearly 110 miles of “superhighways” — a network of what researchers called “the first freeway system in the world.” [Washington Post] https://archive.is/6u7f9
7. Goal-oriented representations in the human hippocampus during planning and navigation https://www.nature.com/articles/s41467-023-35967-6
8. Soft 'e-skin' generates nerve-like impulses that talk to the brain https://medicalxpress.com/news/2023-05-soft-e-skin-generates-nerve-like-impulses.html
9. Olivine weathering can sequester CO2 for as little as $20/T. The cheapest way by far to turn back the clock on CO2. https://worksinprogress.co/issue/olivine-weathering
10. Health officials delayed report linking fluoride to brain harm https://www.salon.com/2023/03/16/health-officials-delayed-report-linking-fluoride-to-brain-harm_partner/
11. “Shoigu's medal "For the return of Crimea" has an interesting story. Its existence was denied until 2018, when Shoigu first wore in it public. The medal dates the start of the operation to seize Crimea as "20.02.14" – while Viktor Yanukovych was still the president of Ukraine.” https://twitter.com/ChrisO_wiki/status/1662937494495657984
12. Total count of visually documented Russian tank losses during their 2022 invasion of Ukraine has reached 2000 https://www.oryxspioenkop.com/2022/02/attack-on-europe-documenting-equipment.html
13. Ukraine launches major drone attack on Moscow https://www.reuters.com/world/europe/drones-hit-several-buildings-moscow-mayor-2023-05-30/ (Videos: 1. https://twitter.com/KevinRothrock/status/1663404037255864320 2. https://twitter.com/guyelster/status/1663405373615665157)
14. “Better footage on how Russians troops were decimated yesterday near Mariupol. It is now reported by multiple sources that a Russian barracks got hit, killing and wounding hundreds of soldiers.” https://twitter.com/NOELreports/status/1663285005089398787
1. Gemini will bring the next big timeline update https://www.lesswrong.com/posts/zhf7LrMmnFbJKqqPL/gemini-will-bring-the-next-big-timeline-update
2. Training LLMs via Simulated Human Societies https://arxiv.org/abs/2305.16960
3. Mindstorms in Natural Language-Based Societies of Mind https://arxiv.org/abs/2305.17066
4. Think Before You Act: Decision Transformers with Internal Working Memory https://arxiv.org/abs/2305.16338
5. LayoutGPT: Outperforms text-to-image models/systems by 20-40% and achieves comparable performance as human users in designing visual layouts for numerical and spatial correctness. https://arxiv.org/abs/2305.15393
6. Under 1,350 square miles of jungle in Guatemala, scientists discovered 417 cities dating back to circa 1,000 B.C. and that are connected by nearly 110 miles of “superhighways” — a network of what researchers called “the first freeway system in the world.” [Washington Post] https://archive.is/6u7f9
7. Goal-oriented representations in the human hippocampus during planning and navigation https://www.nature.com/articles/s41467-023-35967-6
8. Soft 'e-skin' generates nerve-like impulses that talk to the brain https://medicalxpress.com/news/2023-05-soft-e-skin-generates-nerve-like-impulses.html
9. Olivine weathering can sequester CO2 for as little as $20/T. The cheapest way by far to turn back the clock on CO2. https://worksinprogress.co/issue/olivine-weathering
10. Health officials delayed report linking fluoride to brain harm https://www.salon.com/2023/03/16/health-officials-delayed-report-linking-fluoride-to-brain-harm_partner/
11. “Shoigu's medal "For the return of Crimea" has an interesting story. Its existence was denied until 2018, when Shoigu first wore in it public. The medal dates the start of the operation to seize Crimea as "20.02.14" – while Viktor Yanukovych was still the president of Ukraine.” https://twitter.com/ChrisO_wiki/status/1662937494495657984
12. Total count of visually documented Russian tank losses during their 2022 invasion of Ukraine has reached 2000 https://www.oryxspioenkop.com/2022/02/attack-on-europe-documenting-equipment.html
13. Ukraine launches major drone attack on Moscow https://www.reuters.com/world/europe/drones-hit-several-buildings-moscow-mayor-2023-05-30/ (Videos: 1. https://twitter.com/KevinRothrock/status/1663404037255864320 2. https://twitter.com/guyelster/status/1663405373615665157)
14. “Better footage on how Russians troops were decimated yesterday near Mariupol. It is now reported by multiple sources that a Russian barracks got hit, killing and wounding hundreds of soldiers.” https://twitter.com/NOELreports/status/1663285005089398787
😢1
Eliezer Yudkowsky explains why it is very likely that superintelligent AIs won't care about humans AT ALL: https://twitter.com/ESYudkowsky/status/1663313323423825920
I *really* hope he is right. Because the worst thing that can happen is not extinction, but astronomical suffering; being kept alive for some strange reason by a god with an alien psychology.
I *really* hope he is right. Because the worst thing that can happen is not extinction, but astronomical suffering; being kept alive for some strange reason by a god with an alien psychology.
👎7🔥4🤔1
A Who’s Who of the biggest names in AI - including the CEOs of DeepMind, OpenAI and Anthropic, the CTO of Microsoft and many more signed this letter explicitly acknowledging the risk of extinction from AGI.
https://www.safe.ai/statement-on-ai-risk
https://www.safe.ai/statement-on-ai-risk
👍6😴3🤮2
This media is not supported in your browser
VIEW IN TELEGRAM
A remarkable demonstration of how AI will change gaming: this new demo from NVIDIA combines a low latency text to speech engine, a new foundational LLM built for gaming, and an audio to facial expression engine to create truly lifelike NPCs.
👍6
This media is not supported in your browser
VIEW IN TELEGRAM
Ghost in the Minecraft: Generally Capable Agents for Open-World Enviroments via Large Language Models with Text-based Knowledge and Memory https://github.com/OpenGVLab/GITM
- Broad task coverage. All previous agents combined can only achieve 30% completion rate of all items in the Minecraft Overworld technology tree, while GITM is able to unlock 100% of them.
- High success rate. GITM achieves 67.5% success rate on the "ObtainDiamond" task, improving the SOTA (OpenAI's VPT) by +47.5%.
- Excellent training efficiency. OpenAI's VPT needs to be trained for 6,480 GPU days, DeepMind's DreamerV3 needs to be trained for 17 GPU days, while our GITM does not need any GPUs and can be trained in 2 days using only a single CPU node with 32 CPU cores.
"This research shows the potential of LLMs in developing capable agents for handling long-horizon, complex tasks and adapting to uncertainties in open-world environments."
- Broad task coverage. All previous agents combined can only achieve 30% completion rate of all items in the Minecraft Overworld technology tree, while GITM is able to unlock 100% of them.
- High success rate. GITM achieves 67.5% success rate on the "ObtainDiamond" task, improving the SOTA (OpenAI's VPT) by +47.5%.
- Excellent training efficiency. OpenAI's VPT needs to be trained for 6,480 GPU days, DeepMind's DreamerV3 needs to be trained for 17 GPU days, while our GITM does not need any GPUs and can be trained in 2 days using only a single CPU node with 32 CPU cores.
"This research shows the potential of LLMs in developing capable agents for handling long-horizon, complex tasks and adapting to uncertainties in open-world environments."
🆒9
Fewer than half of Gen Z men think “feminism has made America a better place”: https://twitter.com/BradWilcoxIFS/status/1663253365902409752
👍8🤣4😁2