Welcome to another weekly Memia newsletter, your regular scan across the latest developments in AI, emerging tech and humanitys exponentially accelerating future.
(And in particular a warm Kia ora to all of the new readers who have signed up in the last few weeksI hope you enjoy the read!)
The most clicked link inlast weeks newsletter(4% of openers) was NVidia CEO Jensen Huang on stage with his robot buddies. Crazy.
ICYMI
Ive been doing some enjoyable interviews and podcasts in the last couple of weeks since the digital launch of my book Fast Forward Aotearoa:
-
BusinessDesk: Peter Griffin is the first to review the book: Time to switch off strategic cruise control ($walled)
Reid has long argued for better foresight capability in government, a need that grows more urgent by the year. The second half of Fast Forward Aotearoa outlines 10 big-picture missions for the country, which act as worthy stand-ins for the defunct Te Ara Paerangi Future Pathways project, which sought to put in place national research priorities for the country. Underpinning those missions…are 100 technologies we could pursue to help achieve them.
This is where Reid really gets to geek out, taking us through some of the breathtaking technological advances we can take advantage of. Few of them are sure bets, but some of them would deliver disproportionately large benefits to a small nation if executed successfully…
At least Reid is considering the big issues and putting them in context for us as a nation. How many of our politicians are pondering how to fix our outmoded governance systems or how to pay for healthcare for a rapidly ageing population?
As a 21st-century nation-state with democratic values, Aotearoa needs to do more to prepare for future negative eventualities driven by exponential technology, Reid points out. This can be done by investing more in open source technological capabilities, strategic foresight and more advanced democracy systems to achieve greater national agency and resilience, he concluded.
He may be 50 years ahead of most of us in his thinking, but Fast Forward Aotearoa presents a compelling case for why we urgently need to catch up.”
-
I also sat down with Peter Griffin for The Business of Tech podcast, take a listen here:
-
I also joined Bernard Hickey and Peter Bale on the weekly Hoon podcast last Thursday.
The print edition of the book is very nearly ready, youd be amazed how many detailed fixes there are to finish. thank you everyone for your patience aiming to open up pre-orders next week for delivery of the first batch late April. More details asap after Easter!
Brain-computer interface company Neuralink put out a livestream last week with 29-year-old quadriplegic Noland Arbaugh – the first human patient to receive a Neuralink brain implant.
The demo is pretty slick and completely natural hes a really nice guy, too, which obviously helps. With the device he can now move a cursor around his computer screen just by thinking which has improved his independence no end. He shares how Neuralink has enabled him to play Civilization 6 for hours and learn new languages, which were previously challenging due to his physical limitations.
(Also, the bit where he casually switches off the background music is nicely done)
We live in amazing times. More details and a longer interview with Noland here at this Neuralink all-hands meeting.
Whatever one may think of Elon Musk, (in light of the weird in-joke that is the CyberTruck and the ongoing meh at X) theres no denying that hes led an incredible series of technological achievements this year with Tesla Optimus, SpaceX and now Neuralink.
The week’s AI news and updates
Mega Silicon Valley VC firm A16Z put out a new report on Enterprise GenAI use – lots of data and insights in here if kept a bit high level and a subjective sample of only 70 enterprise decision makers.
(Mike Hall for sharing).
The league table of foundation model providers delivers no surprises. OpenAI *so far* out in front, still. (although Anthropic catching up rapidly)
One other interesting data point: the rate at which open source models are being used:
One of the questions I get asked a lot (after how does our organisation get started with AI? (Answer: just frickin START, will you) is the question of timing: should we wait for GPT-5 / after the next model release?
AI oracle Ethan Mollick goes into this conundrum in more detail:
When technology is improving quickly enough, there are types of problems that work just like the interstellar journey, where waiting for technology to get better beats acting immediately. This is called a Wait Calculation the logic also turns out to apply to other problems as well. For example, there are certain classes of mathematical problems where the computations would take decades, so you are better off waiting for Moores Law to kick in, and computers to improve, before even starting to try to solve them.
Which brings us to AI. AI has all the characteristics of a technology that requires a Wait Calculation. It is growing in capabilities at a better-than-exponential pace (though the pace of AI remains hard to measure), and it is capable of doing, or helping with, a wide variety of tasks. If you are planning on writing a novel, or building an AI software solution at your business, or writing an academic paper, or launching a startup should you just wait? I can think of at least two major projects where I should have waited, if I had known how good AI was going to get.
He has a valuable decision framework, more here:
The Lazy Tyranny of the Wait Calculation
Say you wanted to get to our stellar neighbor, Barnards Star as fast as possible. When should you leave? You might be tempted to answer as soon as you can, but you should reconsider. Using todays rocket technology, this six light year journey would take 12,000 years. The catch here is ‘today’s rocket technology.’ A theoretically possible fusion-power
2 months ago 373 likes 36 comments Ethan Mollick
At the beginning of this year, I predicted:
A venture funding crunch will drive consolidation among frontier AI labs and the usual Big Tech giants will start acquiring some of the smaller companies.
This has (sort of) played out twice this week sort of.
Firstly, the AI industry had a sharp intake of breath when Mustafa Suleyman, co-founder and CEO of AI foundation model challenger Inflection AI (and formerly a co-founder of Deepmind), announced that he and his core team were leaving Inflection and moving to Microsoft to run a new consumer division Microsoft AI.
You what?
This is the same company (and founding team) that raised US$1.3Bn in funding from Tier-1 investors (led by Microsoft) less than 9 months ago. At the time, they werent shy about it either:
22,000 H100s.!!! Maybe the GPU farm is worth more than the company these days!
The announcement also included news that Inflection-2.5 would now be hosted on Microsoft Azure for Microsoft customers to use.
(Inflection, who provide the Pi chatbot, released underwhelming benchmarks for its updated Inflection-2.5 AI model a couple of weeks ago (blown out of the water by Anthropics Claude 3). This may have sharpened focus when considering the amount of investment required to get to Inflection 3
The industry seems a bit nonplussed by it all: I cant think of another previous scenario where the CEO and senior technical leadership of a startup are allowed to just jump ship across to a major investor in public
Some news coverage which probes the unanswered questions:
Rumours going around that the full terms of the deal werent out in the open:
sources tell methere’s an undisclosed financial component to Inflection’s licensing its API to Microsoft, providing a path for Inflection investors to be made whole over time. @alexkonrad
Hiring the team rather than acquiring the company is definitely one way around anti-trust regulatory attention
Also of note: LinkedIn founder and major Silicon Valley player Reid Hoffman is also a director of Microsoft he was previously on the board of OpenAI until a year ago, stepped down due his conflicts of interest as a co-founder of Inflection. (!)
Tech industry savant Ate-a-Pi seems to grok what happened better than most (as usual):
Above everything else, this seems another GOAT strategic move by Microsoft CEO Satya Nadella he now has Inflection 2.5 in-house and the team which built it. and access to another 22,000 GPUs. ready to use as leverage when OpenAI open negotiations for their next investment round.
Meanwhile for Mustafa, his new job title may not quite be all its cracked up to be:
In the same week, trouble atmill also for another AI foundation model maker: British firm Stability AI, makers of Stable Diffusion and a pioneer of open-source AI. After smoke signals circulating for months that things werent going so well and challenges with funding the company put out this terse announcement on Saturday:
Earlier today, Emad Mostaque resigned from his role as CEO of Stability AI and from his position on the Board of Directors of the company to pursue decentralized AIWe are actively conducting a search for a permanent CEO to build upon Stability AIs foundation and lead the company into its next phase of growth.
On behalf of the Board of Directors, I want to thank Emad for his leadership and relentless commitment to Stability AI and the open source movement, said Jim OShaughnessy, Chairman of the Board at Stability AI.
Media coverage:
(Lots of easy unstable puns I am no exception)
The plot thickened when Mostaque took to X a few moments later, tweeting:
Not going to beat centralized AI with more centralized AI. All in on #DecentralizedAI Lots more
Also intimating that his shares in Stability still had full board control
The concentration of power in AI is bad for us all I decided to step down to fix this at Stability & elsewhere
(Is this some 12-dimensional-chess mirror move of what happened to Sam Altman at OpenAI last year?)
Anyway, watch this space
After OpenAI researchers blew everyones minds with their Sora text-to-video model a month ago, they put out an update on their work with creative artists and filmmakers:
Since we introduced Sora to the world last month, weve been working with visual artists, designers, creative directors and filmmakers to learn how Sora might aid in their creative process.
While we have many improvements to make to Sora, we’re already getting a glimpse of how the model can help creatives bring ideas to reality.
Another clutch of impressive videos showcased, including this one below, AirHead from Toronto multimedia production company Shy Kids:
As great as Sora is at generating things that appear real – what excites us is its ability to make things that are totally surreal.
Another week with a bumper crop of AI advances
-
OpenInterpreter is aiming to build an open source language model computer. The first device is the US$99 01 Light: kind of a competitor to the Rabbit R1 launched earlier this year at CES which basically provides an always-on voice AI interface with your personal computer in the cloud or back home (A clear vision to provide programmability with AI check out the demo video belowthe founder Killians voice is, er, unexpected).
-
EARTH-2 Also announced at last weeks NVidia GTC but lost in the robot noise NVidia announced its Earth-2 climate digital twin cloud platform for simulating and visualizing weather and climate at unprecedented scale.
To accelerate efforts to combat the $140 billion in economic losses due to extreme weather brought on by climate change
One of the first users of the EARTH-2 APIs is the Taiwan Central Weather Administration, for better detection of typhoon landfall.
-
Cosmopedia New research from Hugging Face on how to create large-scale synthetic data for pre-training:
-
VLOGGER new research from Google, a new AI system that is able to generate lifelike videos of people speaking, gesturing and moving all from only a single still photo.
(Video: Enric Corona et al.)
-
Everyone is an NPC: Games company Ubisoft is developing NPCs that you can have an intelligent conversation with.
-
Synthemol: Another amazing story of applying generative AI for scientific discovery: the WHOs highest priority for a new antibiotic is for treatment of Acinetobacter baumannii. Generative AI to the rescue to design and validate easily synthesizable and structurally novel antibiotics:
Here we introduce SyntheMol, a generative model that designs new compounds, which are easy to synthesize, from a chemical space of nearly 30billion molecules. We apply SyntheMol to design molecules that inhibit the growth of Acinetobacter baumannii, a burdensome Gram-negative bacterial pathogen. We synthesize 58 generated molecules and experimentally validate them, with six structurally novel molecules demonstrating antibacterial activity against A. baumannii and several other phylogenetically diverse bacterial pathogens. This demonstrates the potential of generative artificial intelligence to design structurally novel, synthesizable and effective small-molecule antibiotic candidates from vast chemical spaces, with empirical validation.
Ghost kitchens and AI-generated food:
According to my echo chamber on X, this sounds credible
That was just the week in AI!!
Some more signals from near and far futures…
Two recent pieces gazing into the crystal ball which caught in my net:
-
UK 2020 Options is a project run by UK innovation NGO Nesta looking out one generation ahead:
Children born today will reach adulthood around 2040. What will life in the UK be like for them, according to current trajectories?
The groups latest report The Radical How anticipates an upcoming change of government. Well worth a read:
UK 2040 Options and Public Digital on how to build a mission-oriented government that delivers better outcomes, reduces risk, saves money, and rebuilds trust
The struggles and shortcomings of delivering in government are well rehearsed. Many of the root causes that make it tough have been restated
several times over several decades. But what to do?
.The Radical How is a change of mindset as much as a change in organisation. It promotes methods and processes that have been shown to work, multiple times, at scale. They are the default ways of working for many of the worlds most successful companies.
However, the occasions where they have been deployed are rare in government. These occasions have come about thanks to exceptional leaders, exceptional circumstances, or both.
-
In Australia, the ABC tries to look out 40 years in Life in 2064:
Its 2064. A Queensland great-grandmother, lets call her Lisa, is sweltering in her living room as she waits for a driverless car to take her to hospital. Her robot carer delivers water and medication while a nanobot circulating in Lisas bloodstream sends real-time health data to her doctors.
(I have no idea what may happen in 2064 but my educated guess is that the acceleration of tech means that everything imagined here will be commonplace in 2044, if not 2034)
@levelsio is a software and AI solopreneur who operates at least 6 standalone SaaS and AI businesses (see his profile). These bring in (he claims) over US$200K/month, from which he invests the profits into ETFs which have yielded a further 40% return in 1 year. The business is him, plus one part-time server guy. The rest completely automated:
This raises two alternative scenarios for the future of the software industry:
-
A few giant AI firms create such powerful coding models that creating highly functional software is something any dedicated solopreneur can learn really quickly and manage on their own large software firms become like overburdened bureaucracies vulnerable to small, agile, super-low-cost operators, OR
-
A few giant AI firms create such powerful coding models that creating really functional software APIs, UIs and UX is something the AI just does in real time behind the scenes the global software industry as we know it falls to Microsoft, Google and Apple and their $20/month AI subscriptions.
Which one do you think it will be?
Click Farm Fraud has gone from this
to this in the space of a few years (video via @douglasmun) – putting a bunch of smartphone motherboards inside a single chassis and controlling from one screen:
(Product advertised: CellHasher Smartphone Chassis).
California startup Aethero Space came out of stealth: it is developing radiation-hardened edge computers for on-orbit data processing and eventually even autonomous decision-making.
AI workloads are moving into space Media:
Researchers from from NTT Innovative Devices Corporation in Japan have developed a new indium phosphide-based modulator which achieved an unprecedented net bit rate of 1.8 Tbps over 80-km standard single mode optical fibre
A collaboration between researchers of the University of Tokyo, Johannes Gutenberg University Mainz (JGU) in Germany and Palack University Olomouc in the Czech Republic recently demonstrated a new means of constructing a photonic quantum computer using a laser pulse which can operate at room temperature, which produces a logical qubit with the inherent capacity for error correction. Still a way to go to achieve sufficient stability, though
The new Mori JP Tower in Tokyo is Japans new tallest building reaching a height of 325.2 m. Although its only the 134th tallest building in the world, its a remarkable achievement designed to continue functioning as normal even in a magnitude 9.0 earthquake:
High-strength steel and concrete components are deployed to balance the structures, and vibration-control devices are placed in core areas where vibration during earthquakes can be efficiently reduced, resulting in high resistance to shakingIn addition, large-scale vibration-control devices known as “active mass dampers” will help reduce the swaying that occurs at the tops of buildings during strong winds.”
The development has also received the LEED Platinum green building standard for its sustainability features.
Far smaller, but no less innovative: US construction technology firm Icon Build have prototyped their new Phoenix concrete 3D-printer – which has printed an 8 metres tall 3D-printed multi-storey structure built by a new articulated-arm robot :
we wanted something that was free from architectural constraints that broke us out of rectangular build volumes
Well, quite:
Almost 98% of all global available rooftop surfaces are not used for solar power generation. A study in 2001 found that only 50% of the worlds rooftops would need to be covered with solar panels in order to deliver enough electricity to meet the worlds yearly needs.
Heliasol from German company Heliatek is an innovative flexible thin-foil photovoltaic film which can be retro-fitted onto rooves and the sides of buildings. Its most recent reference site in Korea produces 37.7 kWp (kilowatt-peak).
(Animation: Heliatek)
The product comes with an impressive emissions profile:
Heres the technology being applied to the side of a factory in Nuremberg, Germany:
Envac is a Swedish-HQed company developing urban vacuum-pipe networks for waste disposal: reducing CO2 emissions by up to 90% and removing smells / rubbish bins from cities and other hubs like airports, malls:
With the Envac system, waste handling becomes an integrated part of the urban infrastructure, just like water, sewage, or any other service provided by the city. As the demand for recycling and zero waste increases, so does the need for waste rooms and recycling stations as well as frequent collection and routes for collection vehicles.
In contrast, our waste inlets are automatically emptied and the waste is transported out of the area in underground pipes. This saves bin space inside and outside the buildings and reduces the need for collection routes. The freed-up space can be used for green areas, pedestrian walkways, and bicycle lanes, or smaller businesses, bike repair shops etc.
Heres an artists impression:
And heres a demo video of an installation in an apartment block in Madrid, Spain:
I wonder if the CO2 emissions calculation is completely above board but seems like a good idea for building new housing / city blocks
Switzerland plant-based meat firm Planted has launched a whole-cut vegan steak which uses a proprietary fermentation process which hits flavour, health and climate touchpoints all at the same time.
A quick look up outside the tech world
In the face of attention- and emotion-hijacking mainstream news and social media, Ive been searching for reliable data to back up my gut feeling of greater conflict spreading across the planet.
Two resources to share:
-
Global Conflict Tracker from independent US foreign policy think tank the Council on Foreign Relations. Its grim out there around the world. but apart from Ukraine all conflicts seem to be constrained to the tropics for how long? Most of the dots mapped here are marked Conflict status: Worsening
-
Perplexity AI has introduced a new feature: updated reference answers. I havent checked this answer below in detail but it leverages Wikipedia for its first reference and the summarisation seemed generally correct, I can see functionality like this being very useful to cut through the noise quickly use with EXTREME caution
What were the biggest war hotspots in the last week globally?
-
The climate observatory in Mauna Loa, Hawaii recorded a record annual increase in CO2 concentration, a whopping 3.4 parts per million (ppm) in 2023:
-
The EEA released its latest European climate risk assessment report which doesnt mince its words:
Human-induced climate change is affecting the planet; globally, 2023 was the warmest year on record, and the average global temperature in the 12-month period between February 2023 and January 2024 exceeded pre-industrial levels by 1.5C.
Europe is the fastest-warming continent in the world. Extreme heat, once relatively rare, is becoming more frequent while precipitation patterns are changing. Downpours and other precipitation extremes are increasing in severity, and recent years have seen catastrophic floods in various regions. At the same time, southern Europe can expect considerable declines in overall rainfall and more severe droughts.
These events, combined with environmental and social risk drivers, pose major challenges throughout Europe. Specifically, they compromise food and water security, energy security and financial stability, and the health of the general population and of outdoor workers; in turn, this affects social cohesion and stability. In tandem, climate change is impacting terrestrial, freshwater and marine ecosystems.
Climate change is a risk multiplier that can exacerbate existing risks and crises. Climate risks can cascade from one system or region to another, including from the outside world to Europe. Cascading climate risks can lead to system-wide challenges affecting whole societies, with vulnerable social groups particularly affected. Examples include mega-droughts leading to water and food insecurity, disruptions of critical infrastructure, and threats to financial markets and stability
-
As it becomes apparent that Nations Are Undercounting Emissions, Putting UN Goals at Risk, environmental journalist Andy Revkin surveys the increasing levels of unease around whether current climate models are way out:
Why Recent Chart-Busting Heat Spikes are Such an Unnerving Surprise to Climate Scientists
Updated – Gavin Schmidt, who took the lead at NASAs climate-charting Goddard Institute for Space Studies a decade ago following the retirement of climate pioneer James E. Hansen, has long been a careful data cruncher and communicator. Thats why his Nature commentary
5 days ago 6 likes Andy @Revkin
-
One small piece of good news for the climate: the wind was blowing hard over the weekend and Scotlands energy consumption was almost entirely generated by renewables:
In South Africa, the city of Johannesburg is enduring ongoing water shortages. Mostly this is apparently due to poor maintenance and financial management by the city authorities but future water supply is an issue which is top of mind. Detailed explanation in The Conversation: Johannesburgs water crisis is getting worse why the taps keep running dry in South Africas biggestcity.
(One scenario explored in Kim Stanley Robinsons amazing novel Ministry for The Future is when an entire city in Africa runs out of water as the rivers, dams and wells run dry)
Spains High Court made an order for a temporary ban on the encrypted messaging app Telegram following piracy complaints from media companies – but then promptly suspended the order to more fully investigate the impact of the proposed ban on users following outcry on social media which called the move an attack on free speech.
A few things which Ive been watching and listening to this week
Spatial synth
A whole new class of musical instrument has just arrived heres creative artist Don Allen Stevenson making synth music with his Vision Pro. wow can you imagine the creative altitudes where this can go to…? (You could get an entire orchestra into those fingers..!). Beautiful for its potential….
After over ten years of development, the quirky video game Harold Halibut from Slow Bros studio is finally going to be released on April 16th. The plot and trailer are intriguing, seems very different:
upcoming handmade narrative game about friendship, and life on a city-sized spaceship submerged in an alien ocean. Join Harold as he explores a vibrant retro-future world in his quest to find the true meaning of home when Harold Halibut launches on PC, PlayStation 5, and Xbox Series X/S on April 16, 2024.
Its been 250 years since your home – an ark-like spaceship – fled an Earth on the verge of cold war to find a habitable planet to preserve the human race. You are Harold, a young lab assistant for the ships lead scientist, Jeanne Mareaux. While most of the ships inhabitants have reconciled themselves to a life lived aboard the sunken ship, Mareaux still works tirelessly to find a way for the ship to leave the planet and find a new, dryer home. But the weird, wonderful and diverse people of the FEDORA I keep Harold busy too Until one fateful encounter plunges Harold into a world no one could have imagined – and one that may hold the key to Mareauxs re-launch plans.
Like many of us I suspect most of my leisure time over the last few days has been spent bingeing Netflixs new adaptation of Cixin Lius 3 Body Problem. Epic, cerebral, big concept sci-fi done reasonably well without dumbing things down too much The script and production from the same team who made Game of Thrones and the cast do a pretty good job bringing it to life believably as well. A little bit too, er, *British*, but thats forgivable. In particular the Panama Canal ship-shredding scene is just like I remember it from the book Im enjoying it so far up to S1E6 really need to go back to re-read the novels now!
Finally, I am now a devotee of this British satirical website: Scarfolk Council
Scarfolk is a town in North West England that did not progress beyond 1979. Instead, the entire decade of the 1970s loops ad infinitum. Here in Scarfolk, pagan rituals blend seamlessly with science; hauntology is a compulsory subject at school, and everyone must be in bed by 8pm because they are perpetually running a slight fever. “Visit Scarfolk today. Our number one priority is keeping rabies at bay.” For more information please reread.
(So much of early childhood in there!)
Typical content: the Scarfolk Council Gullibility Campaign 1976:
And thats another week over with all the best for a relaxing Easter holiday weekend more next Wednesday (or before if I get the book to print on time)
Thanks as always to everyone who takes the time toget in touchwith links and feedback.
Namaste
Ben