Memia 2023.03: A day is a long time in generative AI📈🤯// fusion in 5?⚛️// the new Moon rush🌕// getting NeRFy🥽// Atlas gets a grip🦾// immaculate conception👶
Memia 2023.03: A day is a long time in generative AI📈🤯// fusion in 5?⚛️// the new Moon rush🌕// getting NeRFy🥽// Atlas gets a grip🦾// immaculate conception👶
A colossal astronomical tapestry
Welcome to another weekly Memia scan across new emerging tech and thinking about the future. Lots going on as always…!
(ℹ️Reminder: these weekly newsletters are *way* too long for most email clients…it’s recommended to click on the title link above to read online or view in the Substack app.)
Weekly roundup
🔝The most clicked link in last week’s newsletter (only 3% of openers, a broad spread of clicks last week) was the Ring home patrol drone.
✍️Readers’ lives
I received two pieces of reader feedback this week, thought I’d share:
“Had to chuckle reading your newsletter this week. As I went through various AI amazing news - happening so quick, one can not even keep up, and then AR/VR, computer controlled by your tongue, real 3D built houses, bird drones, DNA cutting, rebooting ourselves and never getting old and Quantum money. THEN I GET TO THE HEADLINE THAT STATES "Mind Expanding" .... Wait, what - The above was not in that category, ... Jesus - not sure I dare to read on now - LOL.“
…and, lest I get ahead of myself🙈:
“The newsletter’s great. It’s the only one that I read for ages when I’m on the toilet“
*Wherever* you’re reading this, a big 🙏 for letting me into your inbox every week.
(Reminder if you’re not already a paid subscriber, special offer still available until the end of January):
Also in the past week…
👏Leaving a legacy
Congratulations to outgoing Prime Minister of my country Aotearoa New Zealand on stepping down after 5 years in office — and perfectly executing a transition to new PM Chris ”Chippy” Hipkins, within only 6 days. As the former Covid-19 government minister commented:
“Leadership contests don’t have to be like the Hunger Games“
I jotted down a few thoughts earlier this week in ⏩Fast Forward Aotearoa #25:
“She has definitely spoken to the next generation more than the previous one, leveraging social media and somehow finding time to hold frequent Facebook Live sessions with a generally adoring younger audience…
As a parent of 3 daughters I’m thankful for how she's modelled alternative behaviours to get ahead and demonstrated more about what's possible for them to achieve than any other public figure to date... and had a baby in her first year in office as well! If nothing else, Ardern leaves a massive, door-opening legacy for the next generation.“
Enjoy a well-deserved break, Jacinda. Thank you for your service.
📉📈Smaller, smarter(?) Big Tech
More significant tech layoffs this week…
Reports that Microsoft is cutting up to 11,000 jobs - around 5% of its total global workforce.
Immediately followed by Google parent Alphabet officially announcing 12,000 jobs are to go at its various subsidiaries (~6% of its workforce).
Functions which have reportedly been most affected include Microsoft gaming (Halo) and mixed reality (Hololens) teams and Alphabet’s Waymo (autonomous vehicles) and Google’s Fuschia Operating System teams.
Word on the street is that Microsoft is OUT of mixed reality for good.
In total, ~130,000 people have been dismissed from their jobs at large tech/media companies in the last 12 months while overall US unemployment is just 3.5%. Derek Thompson in The Atlantic investigates: What the Tech and Media Layoffs Are Really Telling Us About the Economy.
…while simultaneously pivoting to huge generative AI investment:
Google has brought back founders Larry and Sergey to firm up their response to ChatGPT, planning up to 20 new AI products including a chat-based search engine.
(See also Perplexity.ai below…)
As trailed last week, Microsoft consummated their multi-billion dollar, multi-year “investment” deal with OpenAI
, although precise terms haven’t been officially released yet (rumoured at over US$10Bn at US$29B valuation).(See Sam Altman interview below on his monetisation thoughts…)
🌡️CO2 Removal - very early
Flagged last week, the first comprehensive State of Carbon Dioxide Removal report was launched:
“Carbon Dioxide Removal (CDR) involves capturing carbon dioxide (CO₂) from the atmosphere and storing it for decades to millennia on land, in the ocean, in geological formations or in products.
Innovation in CDR has expanded substantially, exemplified by R&D, in patents and capacity investment. CDR has been subject of increasing public attention, and the peer-reviewed scientific literature now consists of over 28,000 English-language studies, growing at a faster rate than for climate change as a whole.“
Summary from the authors here on CarbonBrief: The state of ‘carbon dioxide removal’ in seven charts.
🐄💨Danone with methane
Staying with GHGs, massive global dairy company Danone announced that it is targeting a 30% absolute reduction in methane emissions from fresh milk used in its dairy products, the first food company to align targets with the Global Methane Pledge.
Reminder:
“Cattle are the No. 1 agricultural source of greenhouse gases worldwide. Each year, a single cow will belch about 220 pounds of methane. Methane from cattle is shorter lived than carbon dioxide but 28 times more potent in warming the atmosphere”
Over to you, Fonterra…
🎲Ukraine grand game?
Nearly all of the recent mainstream news stories about the Ukraine war are focused US and European weapon contributions to take on Russia’s looming Spring offensive.
On the one hand, Ukraine clearly needs the weapons to stand a chance of defending and regaining their territory. But some Western voices are starting to look ahead on who is set to benefit most from a long and protracted war?
ABC in Australia: As the war rages on and military spending booms, the US arms industry is a big winner in Ukraine
India-based geopolitical writer S.L. Kanthan puts it more provocatively:
[Weak] signals
Signals from near (and ever nearer) futures…
📈🤯A day is a long time in generative AI
(1960s UK prime minister Harold Wilson once remarked that a week is a long time in politics…well…)
Researchers found that ChatGPT performs well enough to pass the US Medical Licensing Exam (USMLE).
GPTZero is a tool which claims to detect AI-generated plagiarism, for example in school essays: Did a Bot Write This?: New Tools Are Already Taking Aim at AI-Generated Work.
There’s clearly going to be a Darwinian evolutionary sprint race in Education for a few years as the inertia of written-exam-based business models push back… but ultimately a pointless exercise as the LLMs get progressively even more sophisticated than they already are. (And tbh, why would anyone need to be tested in a skill that a computer can do in a few seconds? We were here 9 years ago with calculators in classrooms).
Early sightings of a ChatGPT paid tier for US$42/month (pricey but likely worth it for many…and 42 may be a joke during A/B testing…)
Man of the moment, OpenAI CEO Sam Altman discussed all of these questions and others last week in a wide-ranging 35-minute interview with StrictlyVC’s Connie Loizos. Some deep insights on the ChatGPT inside story and future directions.
In this case I recommend watching the video at 2X speed, but alternatively choose your summary:
The Verge: OpenAI CEO Sam Altman on GPT-4: ‘people are begging to be disappointed and they will be’
Or…. I used this new free tool Summarize.tech, which took less than 1 second to generate this summary of the video below: https://www.summarize.tech/www.youtube.com/watch?v=ebjkD1Om4uw
Summary: The speaker in the video discusses the importance of safety in artificial intelligence, and how companies need to be careful when developing such systems. He also talks about the current state of the startup world, and how it has become more difficult to raise capital and hire employees.
00:00:00 The speaker discusses how they've learned to release technology in a responsible way, and how they believe society will need to adjust to the arrival of artificial intelligence.
00:05:00 The video discusses how StrictlyVC is approaching AGI development, and how they plan to disappoint those who are expecting an "ask" from their AI system. StrictlyVC has created a model that allows for multiple viewpoints to be represented, and they are hopeful that this will lead to AGI being democratized.
00:10:00 The speaker discusses the importance of user control and empowerment of AI, and how Microsoft's partnership with OpenAI is a good example of this. He also comments on the recent suspension of a Google employee with experience in responsible AI.
00:15:00 The speaker discusses some of the concerns that educators have about the potential effects of new technologies, including concerns about the impact of generated text technology on homework and the impact of such technologies on student privacy. The speaker also discusses the importance of individual liberty and the need for society to allow for a wide range of expression in order to allow for a lively and democratic culture.
00:20:00 Sam Altman, OpenAI's president and co-founder, discusses the future of artificial intelligence with a guest. He notes that while the best case scenario is unimaginably good, the potential for an accidental misuse of AI is also a serious concern. He advises caution in prematurely declaring victory on the issue, and expresses hope that more work on AI safety and alignment will help us make the technology safer.
00:25:00 Sam Altman, CEO of Y Combinator, discusses the current state of San Francisco and Southern Valley and how technology companies like Google and Facebook have some responsibility for the problems these regions face. He also discusses his thoughts on Ubi and how it is an important but far from sufficient technology for solving the world's problems.
00:30:00 Sam Altman, OpenAI's CEO, discusses the importance of safety in regards to artificial intelligence and the challenges that companies face when trying to create safe systems. He suggests that companies should start their businesses during a time when technology is evolving quickly, as there will be more opportunities for mistakes to be made.
00:35:00 Sam Altman, president of OpenAI, discusses the current state of the startup world and how it has changed in the last few years. He notes that raising capital has become harder, and adds that hiring and rising above the noise are also harder. He recommends entrepreneurs focus on building a valuable product and creating a moat that will protect the company from competitors.
Wow. Not perfect yet, but very close…
One topic which Altman touches on quite a lot is the AI safety / regulation question:
“The best case is so unbelievably good that it's hard for me to even imagine. I can sort of think about what it's like when we make more progress of discovering new knowledge...than humanity has done so far...in a year instead of 70,000….The bad case, and I think this is important to say, is lights out for all of us. I'm more worried about an accidental misuse case in the short term... I think it's impossible to overstate the importance of AI safety and alignment work. I would like to see much more happening”
For this reason, he says that OpenAI may sit on new features for quite a while to ensure that society is ready for them.
AI alignment work is not glamorous: TIME reports that OpenAI used Kenyan workers on less than $US2/hr to label the data to make ChatGPT model less toxic.
Emad Mostaque, founder of OpenAI challenger Stability AI challenges OpenAI’s censorship-based approach as “fundamentally wrong”.
⚛️Fusion in 5?
One remark in the Q&A with Sam Altman related to another of his major projects, nuclear fusion startup Helion Energy. Here’s the transcript:
Q: Fusion…when do you think there will be a commercial planned actually producing electricity economically?
A: yeah I think by like 2028, pending you know good fortune with Regulators we could be plugging them into the grid. I think we'll do, you know, a really great demo well before that like hopefully pretty soon
(An announcement straight out of the Elon Musk gamebook… but maybe more serious?)
I asked Perplexity AI to fact-check, “the world’s first conversational search engine”:
Its answer:
(It’s a pretty compelling research tool and I’ve started using it instead of Google quite a bit this week…)
🌕The new Moon rush
The Economist reports on the race between three private companies to land a craft on the moon for the first time in 2023 and how that raises major questions on the “ownership” of lunar resources:
hakuto-r Mission 1, operated by Japanese firm ispace, is already on its way and is scheduled to land in late April:
Nova-c, by Houston startup Intuitive Machines:
Nova-C, Credit: Intuitive Machines Pittsburgh, Pennsylvania-based Astrobotic Technology’s Peregrine lander:
Peregrine Lander, Credit: Astrobotic Technology
“The Outer Space Treaty of 1967, space law’s foundational text, is showing its age. It dates back to the era when only governments had access to space. And it states that no claims of sovereignty can be made, on the Moon or elsewhere. Efforts to update the treaty to establish rules around resource extraction have run into the lunar regolith. America has refused to sign the Moon Agreement, adopted by 18 countries in 1984, whereas China and Russia have rejected America’s latest proposal, the Artemis accords of 2020.
The debate over who owns the Moon has been the subject of speculation since long before the space age…As the trio of landers now hoping to reach the lunar surface illustrate, cheap rockets and new technology mean that the previously fantastical question of the Moon’s ownership is about to get very real. These craft are the first representatives of a planned flotilla of lunar vehicles, both crewed and uncrewed, publicly and privately funded, that herald a new Moon rush. That brings possibilities, but it also raises tricky questions about the trajectory of humanity’s exploration and exploitation of space.”
🤑China CBDC Smart Contract
Coindesk reports that China’s e-CNY digital Yuan now has programmability - a lottery by e-commerce app Meituan uses a smart contract to automatically distribute a daily lottery prize of 8,888 RMB (US$1,312).
The technical implementation details are thin on the ground right now — last September the Central Bank's Digital Currency Institute announced it was working to increase e-CNY smart-contract functionality.
On the one hand, this isn’t hugely different to automated payments which exist in every banking system around the world already…
But on the other, this promises a step change in the functionality of Fiat money — in theory these Smart Contracts could be built to easily support ever-more complex financial rules… including, as I wrote way back in 2020, redistributive UBI payments.
It also potentially opens up China’s national currency to a far higher risk of hacking.
🥽Getting NeRFy
As Microsoft apparently exited the Metaverse, Apple quietly slipped its timeline for releasing a set of AR glasses due to ongoing technical challenges, with reports that it is instead focusing on its mixed reality headset range, the first high-end developer-focused version of which is anticipated later this year, expected to cost around US$3,000.
Meanwhile the NeRF (Neural Radiance Field) technology which is likely to underpin much of Apple’s mixed reality UX has moved on considerably in the last year:
🦾Atlas gets a grip
Hyundai-owned Boston Dynamics released their latest showcase video, this time showing their humanoid robot Atlas *WITH HANDS*, cooperating with a human (who isn’t pushing it with a stick this time) in a cobot scenario.
These videos from BD have been are always pure entertainment… impressive advances every time, but *still* after 30 years the commercial release timelines for their robotic technology are uncertain. A hard problem to solve.
(Compare with: Tesla’s shamefully awful Optimus robot reveal last year (covered in Memia 2022.39): a trio of T-shirted engineers holding the prototype up and moving its legs to help it walk was top comedy straight out of HBO’s Silicon Valley.
Billion qubit chips
Engineers at Australian quantum computing startup Diraq have discovered a new way of precisely controlling single electrons nestled in quantum dots that run logic gates…bringing closer the reality of achieving billion-qubit quantum chips.
🔋300% more battery
Also over in Australia, researchers led by Dr. Shenlong Zhao from the University of Sydney, has developed new sodium-sulfur battery technology (a type of molten salt that can be extracted from seawater) which has four times the energy capacity of lithium-ion batteries and is much cheaper to produce.
🏍️Just another e-bike
Hopefully the end is near for loud farty-sounding motorbikes and their obnoxious riders…
👶Immaculate conception?
Conception is a US startup focused on turning stem cells into human eggs:
“We are working on a technology that would give women the opportunity to have children well into their forties and fifties, eliminate barriers for couples suffering from infertility, and potentially allow male-male couples to have biological children.
Long term, this technology could be a critical platform allowing for widespread genetic screening of embryos. If proven safe, it could even enable for genetic editing to eliminate and reduce the risk of devastating diseases for future generations“
(Refer also to last year’s Matrix-like concept for EctoLife, “The World’s First Artificial Womb Facility”).
VTOL Mach 0.8
Jetoptera, a Seattle-based arial mobility firm said that their bladeless vertical takeoff and landing (VTOL) aircraft concept has demonstrated the ability to reach speeds of 0.8 Mach (988 kph) during testing: faster than a Boeing Dreamliner and twice as fast as tiltrotor designs. (The propulsion uses a proprietary Fluidic Propulsion System (FPS) … not dissimilar to a Dyson Fan.)
Fantasy cities
This video of 15 new future cities is actually quite an entertaining run through some of the world’s mega city projects (mostly) under construction. Particularly the contrast between the CGI-future-city-architecture-porn and the empty, dusty, unfinished construction sites in the desert.
🐢🛳️Terayacht
Speaking of fantasies…Pangeos is a 660m-wide, 60,000 -person capacity US$8Bn floating mega city concept in the shape of a turtle… whatever.
Mind expanding
Some longer reads I’ve been dipping into this week:
📈Not enough growth…
Eli Dourado in Heretical thoughts on AI looks at Solow’s paradox, namely:
“You can see the computer age everywhere but in the productivity statistics.”
Intuitively, introspectively, the productivity statistics should be through the roof. But they aren’t. That’s why we have economic data, so that we aren’t fooled by our intuitions….markets are not yet expecting any productivity gains from AI. What if AI ends up like the Internet—transformative to our daily lives while somehow not actually delivering major productivity gains?
He then goes through potential AI productivity gains sector-by-sector across the “real” economy: housing, health, energy… The question is will Solow’s paradox continue to apply or will AI actually finally transform these traditionally inert industries?
📉…vs. Degrowth?
Ecological overshoot writer Erik Michaels provides a doom-laden counterpoint analysis in What is degrowth?
“The reality and gravity of the situation we find ourselves in remains grim simply due to the implications of the actual outcome. Mass die-off is assured, the collapse of industrial civilization is assured, and many studies demonstrate that extinction is inevitable.
So, given all of this, and combined with the fact that growth is no longer possible, it should be evident to all that we have no choice but to promote degrowth and condemn growth. Continuing to attempt growth will have the outcome of a steeper collapse which is already in progress. The same is true for continuing technology use and for attempting to maintain civilization.“
The AI Mediated Self
What if you could prime an AI language model with all your childhood notebooks, letters and emails? Would this enable you to accurately simulate what it would be like “to talk to your childhood self, based on real data sources during that time period, vs trying to imagine how your younger self was”. Would this be a wise thing to do…? Eryk Salvaggio investigates the AI-Mediated Self:
🧬Designing DNA with AI
Elliot Hershberg explores a recent preprint paper outlining potential strategies to apply generative AI to design DNA:
“What is the take home message? This study is a compelling example of the immense promise at the intersection of AI and biology. It demonstrates where ML can have the highest impact: as a powerful tool to guide empirical engineering. Complex regulatory DNA sequences were designed purely in the world of bits. When they were manifested in the world of atoms, they behaved as expected.
Let’s extrapolate for a second. What if we developed a model that could not only design cell-type specific expression, but could design sequences that would drive gene expression in response to signals from the cellular environment. What if we put some of these enhancers into an engineered cell therapy that detected disease and expressed antibodies or mRNA therapies and released them to treat the condition?
I, for one, am excited for our sci-fi future.”
Nuggets and gems
A few finds from around the internet this week…
The future arrives, 100 years later
The 64 million sq km question
I wrote last year in Memia 2022.22 about 🌍🌎🌏The 64 million square kilometre question … namely a research study which discovered that the minimum global land area requiring conservation to prevent major biodiversity losses is at least 64 million square kilometres (44% of the world’s land area).
Here’s a beautiful visualisation of the world’s forests and discussion in VisualCapitalist: Mapping the World’s Forests: How Green is Our Globe?
“forests cover 31% of the world’s land surface. They absorb roughly 15.6 billion tonnes of carbon dioxide (CO₂) every year.”
A colossal astronomical tapestry
On Jan 18, astronomers released a gargantuan survey of the galactic plane of the Milky Way. The new dataset contains a staggering 3.32 billion celestial objects — arguably the largest such catalog so far. You can view the entire survey, with the ability to zoom in and out, at the Legacy Survey Viewer.
Thanks for reading! As always, please do reach out and let me know your thoughts on any of the items covered / share some links.
Until next time.
Ben
FT gift link, free for first 3 people to click.