Nature is disappearing🐾 technological metamodernism🔄 drone couture🥽 chopsticks to catch a Starship🚀 slop (noun)🥘 INTELLECT-1🧠 spot on Spotify👨💻 AI for everyone, by everyone🌍 #2024.41
It's all coordination problems
Welcome to this week's Memia scan across emerging tech and the exponentially accelerating future. As always, thanks for being here!
ℹ️PSA: Memia sends *very long emails*, best viewed online or in the Substack app.
Listen to an AI-generated discussion of this week’s newsletter here:
🗞️Weekly roundup
The most clicked link in last week’s newsletter was little-known (until now) Enterprise AI startup Glean, named as a direct competitor in OpenAI’s recent investment round.
🕵️♂️Not Satoshi
According to last week’s much-hyped HBO documentary (which I haven’t managed to watch yet…), early Bitcoin developer Peter Todd was “unmasked” as Satoshi Nakamoto. He is… surprised.
Reddit is unimpressed. Maybe not the best researched doco ever made....
Another documentary from a few years ago on Youtube, Bitcoin - Unmasking Satoshi Nakamoto is more forensic, leaning towards British cryptographer Adam Back - inventor of the original 1990s HashCash and CEO of blockchain technology firm Blockstream. (Earlier this year the full email correspondence between Back and Satoshi Nakamoto became public after being entered into the official court records in the UK… seems to be two different people).
Personally, I still lean towards the Len Sassaman theory, possibly in collaboration with Hal Finney. The mystery endures…
🔔ICYMI
Anthropic CEO Dario Amodei put out a landmark essay last week - Machines of Loving Grace. In it, he articulates why expects “Powerful AI” (his preferred term to “AGI”) within 2-3 years… and the implications are profound (yet subtle).
I pulled together a few notes at the weekend…. (Midjourney came up with this beautiful image - which @SamRag augmented using Magnific…)
🚀🚗🤖The week in Musk-land
Two of Elon Musk’s companies had mixed receptions last week…
🤖We, Robot
Tesla held an invite-only “We, Robot” event at the Warner Bros. studios lot in Burbank, California…
CEO Elon Musk was on hand to unveil two new concept vehicles:
The “Cybercab” - basically a fully autonomous Tesla car without a steering wheel
The “Robovan” - a fully autonomous, er, minibus. (Just don’t try to ride the zero-clearance concept vehicle up a hill or over a speedbump)
(Neither vehicle has a hard timeline to production…Cybercab target is 2026 subject to passing autonomous driving regulation… all pretty vague).
Tesla also showed off the latest demos of their apparently walking, talking, charades-playing humanoid robot Optimus which had all the moves mixing with the crowds… (click for video):
(Obviously) it turned out that the robots were being teleoperated by humans at least some of the time…not exactly transparent of Tesla… and clearly still a long way to go. (Lots of Starlink dishes on the roof implies operators were physically back at Tesla HQ or even further afield…)
But also impressive to have such natural movement in near-real-time. Even without autonomy, the scenario where robots in the US are teleoperated by staff located around the world in low-labour-cost countries would have major implications for immigration flows…
The aesthetics of all three products oh so closely echo the 2004 film “I, Robot” directed by Alex Proyas… which I guess was the point.
So… lots of Musk’s “future visionary” schtick in a transparent attempt yet again to pump retail demand for Tesla stock…. but this time the market saw through it pretty easily, trimming over 10% off the stock price within 24 hours:
Meanwhile no low-cost model announcement which was rumoured beforehand…
🚀Chopsticks to catch a Starship
Only a couple of days later, Musk’s other major company SpaceX successfully completed Starship Flight 5 - catching the giant SuperHeavy Starship booster rocket with 'Chopsticks' on the landing launch tower — on the first attempt! This really is an amazing technological feat (“chopsticks” catch around 7 minutes in):
(The main rocket ditch into the Indian Ocean and exploded…and presumably sank to the bottom of the sea…leaving behind who know what toxic chemicals?)
SpaceX continues to agitate against US regulator FAA to get more launches approved, faster. But are the environmental harms from GHG emissions and other toxic gases and chemicals truly understood…?
🎭Managing Elon
Whatever one thinks of Musk’s erratic public persona — including his unfortunate star jumping at a Donald Trump rally — this (unauthenticated) anecdote on Tumblr may provide an insight into the complexities of “managing Elon” inside his companies… feels genuine enough.
📈The week in AI
The week's AI news and releases
🔬AI takes science Nobels
This year’s Nobel Prizes in Chemistry and Physics both went to groundbreaking achievements in AI, underscoring AI's growing impact on scientific research:
Demis Hassabis and John Jumper of Google DeepMind won the Chemistry Nobel for developing AlphaFold, the AI model that solved the 50-year-old protein folding problem. AlphaFold has now been used by over 2 million scientists for various applications, including vaccine development, with clinical trials for new drugs expected within two years.
Geoff Hinton, the veteran “father of AI”, received the Physics Nobel alongside physicist John Hopfield for their work on neural networks — the foundation of most modern AI.
In this press conference facilitated by University of Toronto where he has taught most of his career, Hinton (a notable AI Doomer) spends much of his time talking about his concerns on existential AI Risks to humanity (including a dig at Sam Altman for OpenAI’s change from mission-driven to profit driven entity…):
He also gives Canadian research funding body Natural Sciences and Engineering Research Council (NSERC) a shout out for:
“funding basic curiosity-driven research”
Lots of countries could do well to follow Canada’s example…
📊Industry news
A simple visual breakdown of OpenAI’s finances… in case you were wondering:
Related: The fastest-depreciating asset in history? @Swyx in Latent Space newsletter comments on Eugene Cheah’s analysis of $2 H100s: How the GPU Bubble Burst:
“Last year, H100s were $8/hr if you could get them. Today, there's 7 different resale markets selling them under $2…Do we yet have an answer to the $600bn question? It is now consensus that the capex on foundation model training is the “fastest depreciating asset in history”, but the jury on GPU infra spend is still out and the GPU Rich Wars are raging. Meanwhile, we know now that frontier labs are spending more on training+inference than they make in revenue, raising $6.6b in the largest venture round of all time while also projecting losses of $14b in 2026. The financial logic requires AGI to parse.”
US Series A investment is dominated by…AI and SaaS:
☢️Google goes nuclear Google signed a deal with nuclear energy company Kairos Power to purchase electricity from its molten fluoride salt small modular reactors (SMRs) to power its data centers. The first SMR is expected to come online by 2030, with a total of 500 megawatts added to the grid through 2035.
Stripe and NVIDIA announced an expanded collaboration where Stripe will leverage NVIDIA's AI platform to improve its fraud detection capabilities and optimise checkout processes, while NVIDIA will use Stripe's platform to enable developers and enterprises to prepay for select NVIDIA cloud services.
(Also this week, Stripe re-introduced crypto payments in USDC after a 6-year hiatus from 2018… another signal of the imminent mainstreaming of crypto?).
Generative AI’s Act o1 Giant VC firm Sequoia published their analysis of the shift towards inference-based scaling now evident from OpenAI’s o1 ("Q*” or “Strawberry”) model.
“A New Scaling Law: The Inference Race is On: The most important insight from the o1 paper is that there’s a new scaling law in town.
Pre-training LLMs follows a well understood scaling law: the more compute and data you spend on pre-training the model, the better it performs.
The o1 paper has opened up an entire new plane for scaling compute: the more inference-time (or “test-time”) compute you give the model, the better it reasons.“
More to the point, if you try to spot patterns in prior software industry growth, with AI to SaaS as SaaS is to plain legacy Software, what’s next…? (Pretty wild extrapolation… but actually… correct?)
🌍 The Mozilla Foundation released a paper outlining its vision for Public AI:
“Making AI work for everyone, by everyone”
Mozilla is releasing a vision for Public AI, a robust ecosystem of initiatives that promote public goods, public orientation and public use throughout every step of AI development and deployment. Read the paper here.
Look around. There are buses driving alongside cars on the road. Some of your packages are delivered by private couriers, others are delivered by the national postal service. You can flip the channel on your TV back and forth between public broadcasting and commercial networks. And when you access the internet, you can choose between a commercial or nonprofit-backed web browser.
Private and public initiatives have existed side by side for a long time. While private innovation often pushes the frontier of what’s possible, public alternatives can make those innovations more accessible and beneficial for everyone. These parallel products and services give people more choices, create market pressure on each other to be more trustworthy and innovative, distribute power across more people and organizations, and create more resilient and healthier economies.
So, where are the public alternatives for AI? They are starting to emerge, with some governments subsidizing access to computational resources, and nonprofit AI labs collectively putting nearly $1 billion into open source AI research and models. These are important steps forward, but they are not enough to create true public alternatives to the results of the hundreds of billions of dollars going into private AI. This status quo means some critical projects — such as using AI to detect illegal mining operations, facilitate deliberative democracy, and match cancer patients to clinical trials — remain under-resourced relative to their potential societal value. In parallel, Big Tech is ramping up efforts to push policymakers to support private AI infrastructure, which could further cement the dominance of just a few companies in creating the future of AI.“ —Mozilla foundation:
Important initiative to follow… how does it get funded though?
⚖️A/B Testing AI regulation
Meta and OpenAI continue to self-restrain releasing their latest models into the EU due to concerns with the much-ce EU AI Act. This is 🔥:
🎭It’s what’s inside
Netflix launched It's What's Inside: The Game" a digital body-swap experience inspired by the thriller film "It's What's Inside." The game allows 3-6 players to virtually swap faces and voices in real-time, with the aim of trying to work out who’s who:
Congratulations to Mike Hall and team for standing up what sounds like a massive project in just a few weeks… (and standing it down again - the GPU budget is so high it was always going to be just a pop-up!)
🥘Slop (noun)
Seeing this more and more… “slop” as a generic term for AI-generated content… for example searching for “baby peacock” on Google apparently gives you this:
📝The Rise of AI-Generated Content in Wikipedia A study by Princeton University researchers reveals a significant increase in AI-generated content on Wikipedia since the release of GPT-3.5. Over 5% of newly created English Wikipedia articles were flagged as AI-generated, compared to less than 1% before GPT-3.5's release.
WikiProject AI Cleanup is a project aiming to protect Wikipedia from AI-generated information… good luck with that.
Follow on reading… Perplexity page Vitalik Buterin on Epistemic technologies
🆕New research and releases
🧮Not so smart after all? Apple released a new benchmark, GSM-Symbolic which claims to reveal significant limitations in the mathematical reasoning capabilities of LLMs. Current LLMs may not be as capable of true reasoning as previously thought.
Impressive demos of the OpenAI Realtime API in action (🎩 Mike Hall for sharing):
OpenAI published their meta prompts for the "prompt optimization" feature they launched at last week’s DevDay to their documentation (again, I am in wonder that the programming language for AI is… clearly written, precise, instructional English):
“Given a task description or existing prompt, produce a detailed system prompt to guide a language model in completing the task effectively. # Guidelines - Understand the Task: Grasp the main objective, goals, requirements, constraints, and expected output. - Minimal Changes: If an existing prompt is provided, improve it only if it's simple. For complex prompts, enhance clarity and add missing elements without altering the original structure. - Reasoning Before Conclusions**: Encourage reasoning steps before any conclusions are reached. ATTENTION! If the user provides examples where the reasoning happens afterward, REVERSE the order! NEVER START EXAMPLES WITH CONCLUSIONS! - Reasoning Order: Call out reasoning portions of the prompt and conclusion parts (specific fields by name). For each, determine the ORDER in which this is done, and whether it needs to be reversed. - Conclusion, classifications, or results should ALWAYS appear last. - Examples: Include high-quality examples if helpful, using placeholders [in brackets] for complex elements. - What kinds of examples may need to be included, how many, and whether they are complex enough to benefit from placeholders. - Clarity and Conciseness: Use clear, specific language. Avoid unnecessary instructions or bland statements. - Formatting…”
OpenAI also released Swarm, an experimental open-source framework for building, orchestrating, and deploying multi-agent systems. Pretty simple at the most base level, but I can imagine how this scales…
Perplexity now draws charts on request:
INTELLECT-1 from Prime Intellect (first covered in Memia 2024.28) is the first-ever decentralized training of a 10B model:
Why is this important? Firstly it promises to upend the hyperscale-monolith-datacenter AI training model, but bigger picture:
F5 TTS - an open source Text-To-Speech synthesizer which takes a short clip of an original voice and can then faithfully recreate that voice speaking for any text. Try it out yourself at the Hugging Face space here. Here’s Ethan Mollick putting it through its paces on his local PC with just a 10 second clip of him speaking. So easy.
Delphi.ai is just one of many “AI Digital Twin” startups aiming to enable professionals, entrepreneurs and “influencers” to scale themselves to have millions of GenAI conversations at once.
My mate Justin Flitter from NewZealand.ai is first-to-market with a Heygen interactive avatar on his website. The experience is pretty asynchronous (responses take around 10 seconds to process during which time Virtual Justin fidgets a little and glances down awkwardly) - but it’s quite possible to have a meaningful - if vanilla - conversation. Also the accent veers off from Kiwi to West Coast US and back again…😅 Nowhere near ChatGPT Advanced Voice Mode in terms of latency or richness of communication… but a signal of what’s coming likely within a year. Try it yourself here:
Adobe announced Firefly AI video model in Beta, “the first designed for commercial use”. Hmmm, translation: our licensing will bleed you…?
Dreamina 2.0 all-in-one AI generator developed by TikTok parent ByteDance is currently in beta for its new version 2.0, with a wide range of features, including image and video generation, AI music creation, storyboarding, motion brush, canvas tools, and more. Only available in China currently.
(Video via @EHuangLu)
AI with emotions all of these faces were generated with another new Chinese AI video generator, Hailuo AI. Prepare to be manipulated… and desensitised to real human emotional expression. Weird times we are entering…
(video via @maxescu)
🎮The week in drones
Can’t help getting ever-more worried about the military trajectory of (swarming) drones…
🐝10,000 drone swarm
A few weeks ago, a swarm of 8,000 drones was the record. Now it’s 10,000. This is a statement of military might:
(video via @RadarHits)
😨Drones used against civilians
Independent report Caolan Rob reports on how Russia is using drones now to target civilians in the Ukrainian city of Kherson:
☠️Slaughterbots by Anduril
If you remember the cautionary tale Slaughterbots from the Future Of Life Institute 6 years ago… you will be facepalming violently as this video came out from US military tech firm Anduril (founded by ex-Oculus founder Palmer Luckey).
Basically a high speed semi-autonomous kill drone… packaged up in a demo video.
Summarising key trends I’m seeing with military drone use:
The boundaries between human-in-the-loop and full autonomous kill commands are blurring almost beyond recognition now in Ukraine. Also the distinction between civilian and military participants is largely ignored by the Russian side (at least).
Autonomous swarm technology now means that these drone swarms can cover far greater areas with far greater precision and intelligence.
Drone countermeasures such as signal jammers and lasers are now turning up in all sorts of places - for example at Friday Prayers in Tehran last week. As always the Darwinian evolutionary struggle continues on a new fitness plane…
Also seeing trends towards miniaturisation… the envisioned mosquito-sized drones from 12 years ago are surely nearly there now. Try to defend against a million-strong swarm of those armed with neurotoxins…
Long term… human bodies are completely unevolved to survive against these new apex predators. Future war protagonists will spend their whole lives living in bunkers wearing a VR headset to teleoperate in the physical world… I could also see these technologies radically accelerating investment into brain uploading… at some point our conscious self is too precious to carry around in a fragile meatsack without redundancy… maybe within ~10 years we will be backed up and distributed across various datacentres around the world?
🥽Drone couture
On the lighter side, the modern combat drone operator goes hard in front of the mirror each morning…
🔮[Weak] signals
Non-AI, non-drone, non-Musk tech signals from near and far futures...
🏝️.io no more
The British government's decision to transfer sovereignty of the Chagos Islands to Mauritius may have unexpected implications for the tech industry: potentially leading to the disappearance of the popular .io (British Indian Ocean Territories) domain suffix.
👨💻Spot on Spotify
Spotify announced it will continue its "work from anywhere" policy (originally put in place in February 2021). Katarina Berg, Spotify's Chief Human Resources Officer is quoted:
"You can't spend a lot of time hiring grownups and then treat them like children…Work is not a place you come to, it's something you do”
Spot on. 🎯
☀️🛰️Aether Flux
Aether Flux is a new startup building space-based solar power with an approach of using many small satellites transmitting power through infrared lasers, supposdely allowing high power output and small ground stations.
⚡Zap Energy
Fusion startup Zap Energy confirmed a US$130 million funding round and reported progress on its Century power plant demo system, which successfully performed 1,000 consecutive plasma tests in under three hours.
📱RedCap
In the US, mobile networks AT&T and T-Mobile are gearing up to launch their Reduced capacity (“RedCap”) 5G IoT devices as soon as the end of this year. RedCap, based on the 3GPP Release 17 specification, is designed specifically for IoT and offers data transfer speeds of 30-80 Mbps, allowing 5G signals to run in a 20 MHz channel instead of the typical 100 MHz. Initial RedCap devices may include wearables and mini router "pucks" that convert 5G signals to Wi-Fi. (🎩 Andrew Leckie for sharing)
⚛️Chiral nanoparticles
A joint research team from Seoul National University and KAIST has developed a visible light communication encryption technology using chiral nanoparticles, offering high security and fast transmission speeds, making it ideal for local communication systems. Chiral nanoparticles, created by twisting crystal structures using biomolecules, act as unclonable keys in the encryption process.
🧠Mind expanding
(Rolling these links back into the weekly newsletter for the time being, or else I’ll never get around to sharing them!)
🔄Technological metamodernism
Due to my recent travels through various timezones I was unable to book onto the recent online 4-week course Technological Metamodernism given by UK-based transdisciplinary technologist Stephen Reid.
Thankfully, due to some generous donations, Stephen has been able to open-source the course notes. Dive in to absorb 107 pages of weaving together ideas from technology, philosophy, design, game theory, mythology, speculative fiction and consciousness studies… many familiar themes and reference points overlapping with Memia — but all synthesised into a far more coherent framework. Awesome, substantial work, I can’t wait to see what develops out of this.
(The full course notes are also available in markdown format for simple importing into the LLM of your choice).
⏳On the nature of time
Polymath Stephen Wolfram discusses his computational perspective On The Nature of Time:
Key points (summarised by Claude Sonnet 3.5):
Time can be viewed as the progressive application of computational rules that determine successive states of the universe.
Computational irreducibility makes time robust, as there's often no way to predict future states without going through each step.
Our experience of time flowing in one direction is due to the interplay between computational irreducibility and our computational boundedness as observers.
At the lowest level, time is multithreaded with many possible paths of history, but observers like us perceive a single thread due to how we sample and aggregate information.
The concept of The Ruliad - the entangled limit of all possible computational processes - provides a framework for understanding time as part of a larger abstract structure
🤯A structure so large…
Our Galaxy Appears To Be Part Of Structure So Large It Challenges Our Models Of Cosmology - according to the new study, our “local” Galaxy Supercluster Laniakea resides within a larger "basin of attraction" (BoA) potentially 10 times its volume.
🧠Inside every one of us…
I can’t locate the origin of these images shared by Massimo… but the dissected human nervous system is quite something, eh?
(In 1925, two medical students in Missouri spent 1,500 hours of surgery completing the first-ever dissection of the human nervous system. Then they mounted the dissected nervous system on a slab of wood, added hundreds of paper labels to the display and exhibited the finished dissection at medical conferences and museums around the US.)
⏳Zeitgeist
Once around the world lightly...
🌠Auroras
A severe sunburst caused major geomagnetic storms last week, with photos of the aurora coming in from temperate latitudes the world over.
No major incidents with electrical infrastructure reported so far.
Here’s a clip of a Coronal Mass Ejection (CME) actually happening, captured by the SDO Space Observatory (not the one this week I don’t think… but spectacular…)
(Video: SDO via @MAstronomers)
🐾Nature is disappearing
Exceptionally grim reading (1). The WWF’s Living Planet Report 2024 chronicles a staggering loss of 73% of wildlife populations in just 50 years… and warns of imminent tipping points:
“The loss and degradation of nature, accelerated by climate change, is taking us towards tipping points. When nature’s systems pass a tipping point, the consequences can be devastating. Sometimes this happens at a local level – like the collapse of fish populations leading to job losses and reduced incomes in coastal communities. But we also risk crossing global tipping points. Doing so could threaten all our food supplies, trigger widespread disasters like fires and flooding, and destabilise economies and societies everywhere.“
🌡️State of the climate
Exceptionally grim reading (2): The 2024 State Of The Climate Report: Perilous times on planet Earth (published in Bioscience journal from Oxford University Press), uses what would previously have been considered extreme language for scientists, now just sounds factual:
“We are on the brink of an irreversible climate disaster. This is a global emergency beyond any doubt. Much of the very fabric of life on Earth is imperiled. We are stepping into a critical and unpredictable new phase of the climate crisis. For many years, scientists, including a group of more than 15,000, have sounded the alarm about the impending dangers of climate change driven by increasing greenhouse gas emissions and ecosystem change... For half a century, global warming has been correctly predicted even before it was observed—and not only by independent academic scientists but also by fossil fuel companies (Supran et al. 2023). Despite these warnings, we are still moving in the wrong direction; fossil fuel emissions have increased to an all-time high, the 3 hottest days ever occurred in July of 2024.. and current policies have us on track for approximately 2.7 degrees Celsius (°C) peak warming by 2100 …. Tragically, we are failing to avoid serious impacts, and we can now only hope to limit the extent of the damage. We are witnessing the grim reality of the forecasts as climate impacts escalate, bringing forth scenes of unprecedented disasters around the world and human and nonhuman suffering. We find ourselves amid an abrupt climate upheaval, a dire situation never before encountered in the annals of human existence. We have now brought the planet into climatic conditions never witnessed by us or our prehistoric relatives within our genus, Homo“
🌀Weather machines
After Helene…Hurricane Milton did its worst…, at least 27 fatalities were attributed to the storm in the US and Mexico, with severe coastal erosion and destruction in Florida, and reports of neighborhoods submerged under floodwaters.The storm left over 3 million residents without power at its peak.
Designed for Resilience: Pearl Homes in Hunters Point, Florida is a pioneering net-zero community featuring 3-storey, hurricane-resistant, energy-efficient houses with solar power and battery storage. The subdivision incorporates flood-resistant street design and buried utility lines. Residents say homes were minimally damaged and they maintained power during the hurricanes.
(Back in 2013, the environmental advocacy group 350.org petitioned the WMO to name hurricanes after climate sceptic politicians.
Meanwhile conspiracy theorists are focusing on supposed manmade “weather control machines” responsible for making the hurricanes (yes, seriously). They’re so close, but so far away…
🏭Cumulative CO2 Emissions
An alternative view of where CO2 emissions have come from historically:
💻❤️Computer dating
When I was at high school in the UK in the 1980s, there was great hilarity when one of our teachers appeared in a newspaper advert for “Computer Dating” having met his wife through a novel matchmaking service. (Poor guy never lived it down…)
Fast forward 35 years:
(This is a huge stat… nearly all of those matches are through algorithmically-driven, for-profit apps — imagine the long-term attack vectors on population dynamics this could entail…!)
🧘Memetic savasana
Finally, just a few eclectic memes catching in my net this week to share…
💯Happy to share…
Former co-CEO of FTX Ryan Salame plays a mean LinkedIn game - posting this on the day he headed to federal prison to serve 7.5 years…
🇺🇸🗳️Stop the steal
Aussia agitprop merchants TheJuiceMedia cover the US elections… oof, if only the US did satire like this.
🎨#b3d
Blender is the leading free and open-source 3D creation suite. #b3d is a hashtag used in the Blender modelling and animation community to show their creations…. this is just one example of where it’s up to…amazing:
(Video: @leossenas)
🧩It’s all coordination problems
The metacrisis, distilled:
🙏🙏🙏 Thanks as always to everyone who takes the time to get in touch with links and feedback.
Namaste
Ben