Silent Revolution/ New Air-Con Tech as the Future of Cool/ What OpenAI Really Wants/ First Human Organ Created Inside an Animal

Silent Revolution. In today’s Exponential View newsletter, Azeem Azhar opens with a focus on Tesla and EVs, and how incumbent manufacturers are struggling to keep pace with Tesla and overall with the transition to EVs.?

He points out the AI, data, and software advantage cumulated by Tesla over the years, which is putting it in an unassailable competitive position (at least for a while). The best example is the latest Full Self-Driving software (v12) which drives a profound revolution in the field, by relying on only 8 onboard cameras and a neural network approach and managing to provide human-like driving performance, based on previous data/driving videos and not specific lines of code telling what to do in each situation.

The video demonstration is very impressive, and I encourage you to look at it. It also provides an excellent overview of technology and performance. As Azeem states in the newsletter “The future value of the auto industry will be defined by the sophisticated software that governs them, such as Tesla’s FSD”.

While I completely agree with him, I think there is another “silent revolution” driven by Tesla, which is as important as the AI one, which provides for massive competitive and cost advantage compared to all the incumbents. It is the revolution that Tesla is driving on the manufacturing front. Without it, Tesla’s margins and position would be extremely weakened. This is, to a certain extent, the forgotten half of Tesla’s edge.

Tesla’s “silent revolution” has three different components:

1.???? Software first approach

2.???? First-principles manufacturing

3.???? Insane pace of innovation cycles?

Software first approach means that all the hardware is designed and optimized to best serve the proprietary software, and not vice versa. It seems a trivial point, but it is not. To understand why, you just need to have a look at this video of Ford’s CEO.

First-principles manufacturing means that it goes back to first principles in designing the manufacturing process, challenging the status quo on every front. A Reuters article this week gives a good overview of what this means.

According to the article, Tesla has developed a new technique to make electric vehicles (EVs) that could significantly reduce production costs. The company has combined several innovations to enable the die casting of almost the entire underbody of an EV in a single piece, instead of using about 400 parts in a conventional vehicle. The know-how is central to Tesla's "unboxed" manufacturing strategy launched by CEO Elon Musk in March, which aims to churn out tens of millions of cheaper EVs in the coming decade while still making a profit.

The article continues with two sources saying that Tesla's previously unreported new design and manufacturing techniques could allow the company to develop a car from scratch in 18 to 24 months, while most rivals take three to four years. The breakthrough Tesla has made involves the design and testing of giant molds for large parts for mass production and how casts can incorporate hollow subframes with internal ribs to cut weight and boost crashworthiness.

The innovations, developed by design and casting specialists in Britain, Germany, Japan, and the United States, involve 3D printing and industrial sand. Until now, automakers have avoided casting ever-bigger structures because creating molds to make parts of 1.5 meters squared or more is expensive and risky.

As you can see the advantage Tesla is building on the manufacturing front is quite remarkable. And it is not going to stop, as they have put in place an insane pace for their innovation cycle for hardware parts, going down to one single day. Yes, you have read it correctly, it is one single day.

It took a while, and many mistakes, but this is where they are today, and this is also what enables first-principles manufacturing. If you are interested in how Tesla does it, I do recommend this podcast and this video, both are very worthwhile listening to and watching.

The reason why I am mentioning all of this is not because I am an “Elon fan” (I am not, even if I have a huge respect for what he has built). The reason for mentioning it is that there are a lot of learnings there that apply well beyond the automotive industry and are in fact extremely relevant for all (manufacturing) industries that need to reinvent themselves as part of the generative industrial revolution. Those key learnings overlap vastly with the three components of Tesla’s silent revolution:

1.???? We need to fundamentally reverse the process, and start from the software, to then design the hardware with the sensing, actuation, and mechanical capabilities dictated by the software, and not vice versa. Ideally, we should start speaking of the “real twin” and not of the “digital twin”

2.???? Because of 1 (but also 3) we need to challenge a lot of paradigms and “manufacturing debt” (i.e. the status quo) we have accumulated over decades, and work to push boundaries and go routes that have not been explored. It might sound like a platitude, and maybe it is. But it is definitely not the reality for incumbents.

3.???? We need to strive for insanely fast innovation cycles also on the hardware side. This last point is going to become even more important with the adoption of generative AI for hardware design (and matter in general), as the Design-Build-Test-Learn cycle will become the overall bottleneck of manufacturing speed and capabilities.

We are currently building Arsenale BioYards upon these core learnings, and I do encourage everybody attempting to re-do our industrial tissue in a generative fashion to consider them.

New Air-Conditioning Technology Could Be the Future of Cool

July 2023 was the hottest month recorded in human history, with record-breaking heat waves worldwide. Studies have found that the heat waves experienced globally this year “were virtually impossible without climate change." The need for air-conditioning has never been more pressing. Yet, relying on AC to make life liveable is only making things worse.

Nicole Miranda, Senior Researcher of Sustainable Cooling at Oxford, says that the global warming induced increase in the use of air conditioning is “not only a vicious cycle, but it’s an accelerating one.” AC is the fastest-growing single source of energy used in buildings. Numerous startups, like Blue Frontier, cSNSAP, and Transaera, aim to reduce the environmental impact of making buildings cooler.

Sneha Sachar, energy efficiency expert at ClimateWorks, says, “Cooling is a multi-faceted challenge. We need a combination of better buildings and cities, better technology, and a better understanding that the true cost of AC extends beyond electric bills.”

News items:

WeatherBench 2: A Benchmark for the Next Generation of Data-Driven Weather Models

The quiet revolution in weather forecasting accuracy that’s been happening for over 70 years “has been tremendously valuable to society, saving lives and providing economic value across many sectors.” According to Google, machine learning innovations like WeatherBench 2 (WB2) will make more accurate weather predictions and create forecasts “in a matter of minutes on inexpensive hardware.”

?

What OpenAI Really Wants

Sam Altman’s recent world tour included visits with Emmanuel Macron and the prime ministers of Spain, the UK, and Poland. World leaders are dying to know how the GenAI tech Altman helped unleash might “usher in a golden age, or consign humans to irrelevance, or worse.”

ChatGPT got the world’s attention (to say the least). Still, for Altman and his OpenAI cohort, “ChatGPT and GPT-4 are merely stepping stones along the way… to building artificial general intelligence (AGI). A concept that’s so far been grounded more in science fiction than science — and to make it safe for humanity.”

OpenAI’s origin story cast it as a “purely nonprofit research operation.” But it’s now “a profit-making entity… valued at almost $30B.” In interviews with OpenAI’s “top brass,” Wired found that all of them drank the AGI Kool-Aid. The prevailing sentiment is, “Why would a nonbeliever want to work here?”

Altman often speaks of OpenAI’s altruistic aims and mission to guide AGI responsibly. He considered “running for governor of California” but instead realized he was in a position to “lead a company that would change humanity itself.”

“In eight short years [OpenAI] has gone from a floundering bunch of researchers to a Promethean behemoth that has changed the world.” Is that a good thing? Wired interviews OpenAI power players and critics to find out.

News items:

Apple Is Reportedly Spending ‘Millions of Dollars a Day’ Training AI

While most Apple-watchers focus on the recent iPhone 15 launch, the company is quietly “investing millions of dollars a day into AI.” According to insiders, Apple is developing conversational (chatbot) and image generation AI models as well as an LLM, Ajax GPT, trained on “more than 200 billion parameters” and “more powerful than ChatGPT.”

?

The First Human Organ Created Inside an Animal Opens the Door to Manufacturing ‘Spare Parts’ for People

Chinese researchers have “successfully generated a blueprint of a human organ in another animal for the first time.” By inserting human cells into pig embryos “genetically modified… not to develop… porcine kidneys,” the study shows that “hybrid (human and pig)… chimera organisms” can produce half-human kidney cells in 28 days or less.

Despite 150K organs being transplanted worldwide yearly, 100,000 people in the US are waiting for lifesaving transplants — 17 of them die on average daily. Human-pig chimera organs may seem like a quick fix, but significant ethical concerns exist. Colonization of a pig embryo’s brain has actually occurred. Nephrologist Rafael Matesanz, the founder and former director of Spain’s National Transplant Organization believes “The major risk is for the cells to go to the central nervous system and produce a human-pig. Or for them to go to the reproductive system, [which poses the same risk].”

On the flip side, Josep Maria Campistol, General Director of the Hospital Clínic de Barcelona, says that pig/human chimeras could provide “an inexhaustible source of organs, [offering] the possibility of generating specific, personalized human organs for certain patients… I am convinced that, in the near future, we will be able to regenerate chronically diseased kidneys, livers, and hearts to fully or partially restore their function and avoid transplantation.”

News items:

Augmented-reality fighter pilot training system aces first flight tests, Boeing says

Boeing and Red 6 have successfully tested an augmented-reality system in a TA-4J Skyhawk tactical aircraft, with plans to install it on a T-7 advanced training jet. This system is designed to provide pilots with a realistic training environment while minimizing the risks of getting hurt and is expected to revolutionize fighter pilot training for an entire generation.

?

Waiting for AI’s Pull-To-Refresh Moment

The “pull-to-refresh” UX feature will be familiar to any iPhone user. Yet the interface function originated with an indie Twitter client called Tweetie, created by software developer Loren Brichter and subsequently acquired by Apple in 2010. The “mobile era” was a “major paradigm shift in technology,” as were “the internet and microprocessors.” AI looks to be the latest “reset moment that equalizes a lot of tech for one reason.”

Research engineer at Notion, Linus Lee, describes it like this: “In the aftermath of new foundational technology emerging, there is often a period of a few years [when] winners in the market get to decide what the interface paradigm for the technology is.” AI is still waiting for its pull-to-refresh moment — and “the inventor may not come from a big company (and likely won't), it can come from anywhere like Tweetie.”

News items:

Shining a Light on the Digital Dark Age

With cloud storage and hardware backup readily available, a “false sense of security” surrounds “digitized documents.” But unlike the Dead Sea Scrolls, which remain legible “nearly two millennia after their creation… most digital information will be lost in just a few decades… without constant maintenance and management.” Three factors: “hardware longevity, format accessibility, and comprehensibility” could lead to a digital dark age where link rot rules and irreversible data loss are commonplace. This article surveys efforts by the Rosetta Project, the Internet Archive, and others to “secure our digital information so that it survives for generations.”

Should the US Implement a ‘Robot Tax’?

Predictions of a “robopocalypse.” and headlines like “AI Is Coming for Your Job” are needlessly sensationalist and distract us from having a “nuanced conversation” about how automation and AI will “impact human jobs.” Whether the impact is negative or positive, “the workforce of the future will be different, and robotics will almost certainly be a primary driver of that change.”

What — if anything — should be done to address the plight of workers negatively affected by increasing automation? A [robot tax](https://www.brookings.edu/articles/tax-not-the-robots/?), supported in theory by people like Bernie Sanders and Bill Gates, “would disincentivize firms from replacing workers with robots, thereby maintaining human employment. [If] the replacement were made anyway, the tax would generate revenues for the government [to cover] loss of revenue from payroll taxes.” That definition of a robot tax doesn’t directly address the “potential human toll” of jobs lost to automation. A tax could also provide funding to “upskill” workers displaced by robots. And then there’s the argument that automation will actually “create more jobs in the long run.”

News items:

Chinese Scientists Have Developed a New Gene-Editing Tool That Doesn’t Use CRISPR

CyDENT, a new gene-editing tool developed by scientists in China, is “more efficient” than CRISPR, could allow for more precision, and prevent “undesirable mutations.” The discovery of CyDENT coincides with potential US biotech sanctions against China. Export restrictions on biotech could slow down CyDENT’s development “since it does partly rely on imports.” However, study co-author and co-founder of Qi Biodesign, Kevin Zhao, says China is “definitely trying to build out its biotech potential, especially starting at what is needed in research.”