The Roaring 20s/ Carbon-Negative Concrete/The Future of AI/Nobel Prize for Vaccine Logistics?/ Teaching Robots to Dance/ Future of Human Reproduction

The roaring 20s? Towards a new era of innovation. Why a dawn of technological optimism is breaking. The 2010s were marked by pessimism about innovation. That is giving way to hope. These are the headlines in this week’s Economist. And I wholeheartedly agree.

I have been supporting this view for a while now and have two twin reports coming out on January 28th on “Deep Tech and the Great Wave of Innovation” and “Nature Co-Design: A Revolution in the Making” on exactly this topic. The reports are written in collaboration with my friends at Hello Tomorrow and outline the “why”, the “why now” and the “how” of this new era of innovation. I will be sharing them here in two weeks. A quick preview here, putting the current wave of innovation in context:

“The first wave of innovation started with the initial industrial revolutions, laying the foundations of modern industry with breakthrough advancements such as the Haber-Bosch process for making ammonia and Bessemer process for manufacturing steel.

Following World War Two, the second wave of modern business innovation—the information revolution—gave birth to large-company R&D, particularly in the ICT and pharma sectors. Bell Labs, IBM, and Xerox PARC became household names and Nobel Prize workshops. Merck alone launched seven major new drugs in the 1980s. 

In the third wave, the digital revolution, two guys in a garage (or a Harvard dorm room) led the innovation charge, which resulted in the rise of Silicon Valley and, later, China’s Gold Coast as massive global centers of computing and communications technology and economic growth. At the same time, the new field of biotech, also driven by entrepreneurs, fueled much of the innovation in pharmaceuticals.

The wave now taking shape is developing around a new model and promises to radically broaden and deepen innovation in every business sector. It is enabled by crumbling barriers to innovation and powered by the convergence of approaches and technologies”. More on this on January 28th and in Vol. #47!

How We Could Make Carbon-Negative Concrete

Concrete is the second-most used material on earth, behind water. To produce concrete, you mix stone, cement, water, and other ingredients... and then you wait until it "cures". Cement is the glue that holds concrete together, but to make it, you burn limestone with other ingredients, releasing CO2. For every ton of cement manufactured, nearly a ton of CO2 is released: around 8% of global CO2 emissions come from the manufacture of cement.

Solidia Technologies is producing what it calls carbon-negative concrete. It uses the same process for making the cement, but reduces the use of limestone and uses a lower temperature for heating ingredients. Solidia then cures concrete using CO2 captured from industrial facilities, instead of water. During curing, the chemicals in the cement break the CO2 up, capturing the carbon and using it to make limestone. CEO Tom Schuler explains the process and how it can cut emissions by up to 70% in a new TED talk.

News items:

A Newfound Source of Cellular Order in the Chemistry of Life

Inside cells, droplets called condensates merge, divide, and dissolve. Their dance may regulate vital processes.

This Avocado Armchair Could Be the Future of AI

If you think about it, our notion of an artificial intelligent agent is pretty antiquated. Our ultimate measure of machine intelligence - the Turing test - is decades old. VP and head scientist at Amazon Alexa Rohit Prasad recently wrote a great piece on why we should move past Turing's imitation game, given the insights we've gleaned from the past decade of AI. Prasad's argument is that the Turing test doesn't measure the usefulness of AI in an everyday life. (Traditional AI benchmarks in general are problematic. The anonymously submitted Large image datasets: A pyrrhic win for computer vision? closely examined the potential societal harm caused by large, unvetted image datasets, and definitely worth a read.)

OpenAI's unveiling of the DALL·E and CLIP models last week offers a possible alternative to assessing a machine's usefulness in everyday life. DALL·E can generate images from a text prompt, while CLIP is able to accurately match captions with images. The company's work on both models are part of what it calls multimodal neural networks: "A long-term objective of artificial intelligence is to build 'multimodal' neural networks—AI systems that learn about concepts in several modalities, primarily the textual and visual domains, in order to better understand the world. "

Learning the world through two modalities - text and images - is a step closer to the multisensory approach humans use to understand their environments. Which is to say, multimodal neural networks may be a more practical approach to intelligent agents, than say, a chatbot that can fool us for a few minutes. Mark Riedl, NLP researcher at the Georgia Institute of Technology in Atlanta, recently told MIT Tech Review that you don't have to measure DALL·E's intelligence by fooling people into thinking it's human: "DALL·E is the kind of system that Riedl imagined submitting to the Lovelace 2.0 test, a thought experiment that he came up with in 2014. The test is meant to replace the Turing test as a benchmark for measuring artificial intelligence. It assumes that one mark of intelligence is the ability to blend concepts in creative ways."

News items:

Notes on Technology in the 2020s

As we start a new decade, it’s a good time to reflect on expectations for the next 10 years...

IBM Is Using Light, Instead of Electricity, to Create Ultra-Fast Computing

IBM researchers have unveiled a new approach that could mean big changes for deep-learning applications: processors that perform computations entirely with light, rather than electricity.

There Should Be a Nobel Prize for Vaccine Logistics

"Frederick Banting and John Macleod won a Nobel Prize for their discovery of insulin, but there is no Nobel Prize for the many private companies and government regulators who ensure the cold chain actually delivers insulin to those in need. Maybe there should be."

Pathologist Benjamin Mazer considers the much more banal and complex side of scientific breakthroughs: distribution. And the logistics-nightmare du jour is the chaotic and highly criticized COVID vaccine rollout. For Mazer, the intense focus on the botched rollout partly spells a lack of compassion for people just doing their job: "The fact that there’s a caring doctor behind every Covid test doesn’t seem to register. I don’t even try to explain anymore that. despite all of our Covid testing failures, most have not been unique to the US." But there's more to Mazer's argument than personal compassion: "How we talk about our institutions defines them. Whether they seem trustworthy or dishonest, efficient or sluggish, powerful or impotent has as much to do with the public narrative as their actual results." Mazer concludes that we should set reasonable expectations for 2021:

"Setting these in the face of intense, justified emotion is hard. But here’s where we might start: by valuing transparency over perfection, improvement over denial, iteration over omnipotence, and access over perfect equity. We will not administer the vaccine in the exact order that everyone just now personally decided was most fair. Our institutions will fail us this year—as they fail us every year—in ways large and small, incompetent and corrupt. But they will also accomplish tiny miracles."

News items:

In 2023, You Won’t Be Able to Fly Most Drones in the US Without Broadcasting Your Location

The Federal Aviation Administration (FAA) just issued the single biggest set of changes to US drone law since the agency first took an interest in the technology.

How Boston Dynamics Taught Its Robots to Dance

"When you take elite athletes and you try to do what they do but with a robot, it’s a hard problem. It’s humbling." - Aaron Saunders, Boston Dynamics' VP of Engineering

Right before the New Year, Boston Dynamics made its robots dance. The video has already reached 25M views, and for good reason. We've seen Atlas jump over obstacles and traverse rough terrain, but the two humanoids in the company's newest video have never moved so gracefully. IEEE Spectrum asked Aaron Saunders, Boston Dynamics' VP of Engineering, what went into the performance.

Boston Dynamics hired dancers and a choreographer to compose an initial routine, then translated the human movements into something the Atlas robots could actually do. "To do that, we used simulation to rapidly iterate through movement concepts while soliciting feedback from the choreographer to reach behaviors that Atlas had the strength and speed to execute," said Saunders. The team created ad-hoc tools to program the robots' dance moves, a "pipeline that lets you take a diverse set of motions... and push them through and onto the robot."

Saunders and the team even upgraded Atlas's hardware to have it match the strength and speed of dancers: "Dance might be the highest power thing we’ve done to date—even though you might think parkour looks way more explosive, the amount of motion and speed that you have in dance is incredible." It's interesting how the experience has opened roboticists' eyes to the world of dance, meanwhile dancers are having eye-opening moments with advanced robots.

News items:

How the New Virtual Reality ‘Star Wars’ Experience Gets Us Closer to a Theme Park ‘Metaverse’

Star Wars: Tales From the Galaxy’s Edge is a VR extension of the Star Wars-themed lands at Disney theme parks.

The Uncertain Future of Human Reproduction

New data from the CDC points to the lowest rate of births since 1984: "The U.S. had just 58.2 births per 1,000 women aged 15 to 44 in 2019, a 1% decline from the previous year." The US also experienced its lowest population growth rate since 1900. But births are actually increasing among women over 40 thanks to better assisted reproductive technology (ART). Axios gives us a peek at the ethically ambiguous future of ART:

●      "Prospective parents can already employ preimplantation genetic testing (PGD) to screen IVF-created embryos for hundreds of possible genetic abnormalities."

●      "In his fascinating book "The End of Sex," the bioethicist Hank Greely projects a near-future of what he calls "easy PGD," where artificial reproduction — including the ability to screen and select the "best" embryos — essentially displaces making babies the old-fashioned way."

●      "As stem cell science improves, Greely predicts parents of the future won't even need to harvest eggs. Instead, skin cells could be reprogrammed to become egg or sperm cell."

●      "In such a future, LGBT couples could produce children that share the genes of both partners, and single parents could create a child on their own, by generating both the egg and sperm cells from their own bodies."

News items:

Anti-Aging: State of the Art

Today, there are over 130 longevity biotechnology companies and over 50 anti-aging drugs in clinical trials in humans.