SynBio(Beta)/ $1.1T Spent on Climate Technology/ Using GPT to Passively Read People’s Thoughts/ Will a Chatbot Write the Next Succession?

SynBio(Beta). May and June are conference months, at least for me. And, while I mentioned the House of Beautiful Business last week, I am also looking forward to the upcoming SynBioBeta conference, THE synthetic biology conference, which will be held in Oakland 23-25 May. John Cumbers, the founder of the event, has done an incredible job in creating a very strong community, and I am truly looking forward to meeting the community and discussing the future (and the present) of synthetic biology at the event. If you happen to attend, let me know, would be great to meet there. John has put together an amazing roster of speakers, which you can see here, and I am honored to be part of it.

John is also the author of one of the articles featured in this edition of the Antidisciplinarian, about what happens when deep tech meets synthetic biology. And, while he is right in pointing to the need to have more deep tech in synthetic biology, according to my definition of deep tech, synthetic biology IS deep tech. The paradox is easily explained, pointing to where I think the industry should develop.

As I wrote in the report, for me deep tech is an approach to innovation and not necessarily a specific technology or a specific industry. In the same way, technologies are not disruptive per se, as it is the use you make out of them that makes them disruptive.

In essence, because of the lower barriers to innovation, the deep tech approach to innovation enables fundamental (not incremental) innovation thanks to two convergences. On one side the convergence of approaches, i.e. science, engineering, and design (in the form of problem-solving), and on the other side through the convergence of the capability to manipulate matter and energy, the increased computation and cognition power and the capabilities to sense and actuate (hence the three sections of the Antidisciplinarian).

Because of that, i.e. the two convergences, it is possible to generate a much much broader option space, which can be navigated through problem orientation, leading to fundamentally (i.e. orders of magnitude and not percentages) better solutions. 

But, with deep tech, not only the option space is much broader, but the velocity by which it can be explored is also fundamentally reduced, by the application of the so-called Design-Build-Test-Learn cycle, which can be run faster and faster as we can design much more (thanks to GenerativeAi), build much faster (additive manufacturing, DNA synthesis…) and test much more (high throughput screening) generating new insights that lead to better designs and so on.

Reading the above, one would think that my definition of deep tech, almost 100% overlaps with synthetic biology, and the way it is practiced.

The truth is, in my opinion, that while parts of synthetic biology ARE deep tech already, some others are not, yet. Typically, everything that is in the scientific space (i.e. “upstream”), and around protein and molecule, and partially organisms design, is certainly “deep tech” already, ticking all the boxes of my definition.

That said, there is one part of it, typically the industrial one, that is still far from it. And one of the biggest bottlenecks of the industry is exactly the translation of the scientific advancements into an industrial setting (and NOT capacity, as many profess). If we want to solve also the industrial part of synthetic biology, it will have to happen according to the deep tech approach, particularly if we want to reap the rewards of fundamental innovation on that side of the industry as well. 

So, to go back to John’s article (featured below), we can certainly need more deep tech in synthetic biology, particularly downstream, on the industrial side, where deep tech should eventually meet synbio.

This is something you will soon hear more from me about it, as I am working to address this very issue…

How the World Is Spending $1.1 Trillion on Climate Technology

Where’s the “big money” going? According to Bloomberg, the world’s “three biggest economies” and “scores of ambitious VCs” are flocking to startups promising to help make the world carbon-free. The “shift from the world of software into the actual world” led to $1.1T of “global Investment in the energy transition sector” in 2022.

“Horizon-scanning investors” are looking to “[do] deals that push forward the frontiers of climate tech.” Despite “60% of net-zero pathway technologies [not] yet [being] commercially viable,” government initiatives like the IRA have helped “throw off a gazillion investment opportunities.” Bloomberg identifies the climate tech sectors currently receiving the most funding, including energy storage ($18.4B), carbon markets ($5.4B), and agritech ($9.5B).

News items:

Spurred by New Regulations, the Race Is on for Low-Cost PFAS Sensors

PFAS substances - harmful industrial byproducts often called “forever chemicals” - are “widespread” in US drinking water, soil, air, and wildlife. And in Americans. A CDC study “estimated that PFAS chemicals could be detected in the blood of 98% of the US population.” A new EPA proposal proposing stricter limits on six PFAS in drinking water could require greater access to portable tools “that can detect ultralow levels of PFAS in water.” Here’s an overview of current research that could help provide a solution.

cientists Use GPT AI to Passively Read People's Thoughts in Breakthrough

Last week, it was images. Now, researchers using fMRI and an “AI transformer similar to ChatGPT” can non-invasively “interpret the gist of stories that human subjects watched or listened to — or even simply imagined.” By translating brain scans into “continuous language,” the “decoder” can “read peoples’ minds with unprecedented efficacy.”

As a caveat, the study states that “subject cooperation is currently required both to train and to apply the decoder… [But,] future developments might enable decoders to bypass these requirements… and [be used] for malicious purposes.”

News items:

How Microsoft’s Bing Chatbot Came to Be—and Where It’s Going Next

By integrating GPT into its also-ran search engine, Microsoft quickly transformed Bing from a bit of a joke into a “trendsetter.” Since Bing’s chatbot launched in February, “millions of people spanning 169 countries have used it for over 100M conversations.” By providing a “decision engine” instead of a “list of links,” Microsoft has “a rare opportunity to reimagine what web search can be.” What’s next in the company’s crusade to convert Google Search’s 1B+ daily users using into Bing believers?

What Happens When Deep Tech Meets Synthetic Biology

Former Stanford professor and CTO at imec, Peter Peumans, thinks that the “convergence” of deep tech and synthetic biology will “enable the next generation of life science technology and health technology.” Part of imec’s mission is to give “decades-old” tools an “urgent update” in “domains like next-gen sequencing, diagnostics, manufacturing of advanced cell and gene therapies, and synthetic biology.”

Deep tech investment has more than quadrupled from $15B in 2016 to over $60B in 2020. “Out of the box” thinkers like Peumans believe that “leveraging” existing deep tech platforms and applying them to “seemingly unrelated fields” is like “standing on the shoulders of giants: you can get a lot further in your ability to actually impact people's lives.”

News items:

Daimler Is Setting up a $650 Million Charging Network for Commercial EVs

Since 2020, the number of publicly available EV charging ports has increased by at least 40%. There are now over 130,000 public chargers across the US. “But this public infrastructure is almost entirely designed for light-duty vehicles.” Greenlane — a new $650M JV between Daimler, NextEra, and BlackRock — plans to build a charging network “designed specifically for medium- and heavy-duty commercial EVs.”

"Greenlane is designed to begin to tackle one of the greatest hurdles to the trucking industry's decarbonization — infrastructure," said  John O'Leary, President and CEO at Daimler Trucks North America.

Will a Chatbot Write the Next ‘Succession’?

The union representing US TV and movie writers just announced it’s going on strike. Calling the action an “existential fight to ensure the future of the profession of writing,“ WGA members are putting their “pencils down” after six weeks of failed negotiations with movie studios and streamers like Netflix. Despite record studio profits, “writers’ pay has never been lower.”

Much of the WGA’s  Pattern of Demands addresses inadequate compensation. But hidden in the list is a call to “regulate [the] use of material produced using AI or similar technologies.” The NY Times says, “Add screenwriters to the mix of computer programmers, marketing copywriters, travel advisers, lawyers, and comic illustrators suddenly alarmed by the rising prowess of GenAI.”

News items:

An Uncanny and Soulless Realm: A.I. Imagery Being Dubbed “Post-Photography”

Is this photo real? Or is it GenAI?

Thanks to GenAI, we’re now living in a #post-photography era. But the term is “extremely disrespectful to photographers who have honed their craft through years of hard work and dedication” to develop “the visual language… these algorithms draw upon. By differentiating AI-generated images from traditional photography… and recognizing them as a separate and distinct art form… we can ensure that the art of photography retains the respect and admiration it deserves.”

Geoffrey Hinton Tells Us Why He’s Now Scared of the Tech He Helped Build

“Godfather of AI” Geoffrey Hinton is sounding the alarm about the AI tech he helped create. The Turing Award-winning, deep learning pioneer recently dropped the “bombshell” announcement that he’s quitting Google “to talk about the dangers of AI” without having to consider its impact on his former employer. “As long as I’m paid by Google, I can’t do that,” Hinton says.

LLMs, like GPT-4, are “on track” to exceed Hinton’s expectations of how smart machines can get. “For 40 years,” Hinton dismissed “artificial neural networks” as pale imitations of “biological ones.” Not anymore. By “mimicking” the human brain, we may have “come up with something better.” According to Hinton: “It’s scary when you see that. It’s a sudden flip… GPT-4 knows hundreds of times more than any one person does. So maybe it’s actually got a much better learning algorithm than us.”

When asked about LLMs’ propensity to “confabulate.” Hinton sees “bullshitting as a feature, not a bug,” proving that LLMs “are doing something just like people.” Additionally, AI is “very close to being more intelligent than us now and… will be much more intelligent than us in the future,” Hinton says. “How do we survive that?”

Meta’s Chief AI Scientist, Yann Lecun, Hinton’s Turing award corecipient, concurs that “machines will become smarter than humans… It’s a question of when and how, not... if.”…” But he “completely disagrees… that machines will dominate humans simply because they are smarter, let alone destroy humans.”

Hinton’s advice? “Enjoy yourself, because you may not have long left.”

News items:

Robot Dogs Deployed in New York Building Collapse Revive Surveillance Fears

Who let the digidogs out? NYC Mayor Eric Adams declared the Boston Dynamics quadrupedal robots “out of the pound” in April, and the chrome canines were successfully deployed in a recent building collapse. We’re not exactly in Black Mirror territory (yet), but civil liberties advocates worry about the robots’ “capacity to engage in massive surveillance and routinely collect massive amounts of private personal data on millions of New Yorkers.”