HoffaGPT - Lux Capital

Is the future of labor organizing human or machine?

In his risk appendix to Lux’s latest quarterly letter, Josh Wolfe darkly prognosticated that labor unrest will be a significant risk in 2023. Quoting (and keeping the, ahem, extensive Josh formatting):

Let’s riff on that to imagine an improbable but possible contagion few are talking about: a labor contagion. A curious stat caught our attention with important implications for the shape of the future. There’s record high and growing support for unions today (70% of surveyed Americans) yet record low percentage of U.S. workers actually members of one (10% versus 20% 40 years ago). Consider also in 1983 at peak union membership, it was also the peak for interest rates.Might four decades of declining interest rates have coincided with four decades of declining union membership, and as rates rise, might this revert? Might a hard landing, downturn and recession ahead galvanize workers and catalyze a resurgent labor movement? Historically when unemployment and economic stress rose, more resilient industry sectors with more constant demand have been consumer staples, utilities (electric, water, gas), telecom and healthcare. If we see a mass labor union movement fighting for better pay and benefits, healthcare and related services especially may benefit. Let’s continue the thought experiment with an eye towards disrupting daily routines. What would happen if nurses, hospital staff or healthcare providers, transit workers, electricians, teachers, ironworkers, truckers, warehouse workers, police, fire, city workers, and more went on strike? Technological disruption was the theme of the past two decades, but what if labor disruption becomes the theme of our coming era of unrest? Our Covid-19 and banking crises remind us how interconnected and vulnerable we are and how swift social media can exacerbate panic — turning a placid and pacific public into a forceful, furious and ferocious one. As we write, writers are striking in Hollywood, slowing the flood of media Hollywood content from blockbuster movies to late-night talk shows. That’s no laughing matter as labor dissent is a contagion we have not seen in a while. The issue at hand is mostly an economic one regarding participating in streaming revenue, but the West and East Coast writer unions are also focused right now on banning the use of AI in screenwriting. Will a renegade showrunner break the picket line and use ChatGPT and AI script writers to make a point? Are writers the 21st century version of the Luddites of 1800s England or merely the vanguard for putting AI into its proper place? More broadly, will the ubiquity of AI cause increased labor unrest?

On Thursday, it got a bit more meta as Bloomberg reported that AI contractors at Google are attempting to unionize. In their letter to management, the movement wrote that “Instead of recognizing us as highly specialized workers worthy of respect, we are treated as resources for Google to exhaust, despite our expertise and how much genuine care we put into this complex work.” That work reportedly included improving the prompts of Google's Bard chatbot as well as screening for violent and sexual conversations inputted by users.

I read that story just as I saw “Agent-based workflows and waterfall effects” from Matthew Lynley’s thoughtful new ML infra-focused newsletter Supervised. Lynley describes the coming transformation of the workplace:

One of the fastest-growing buzzwords (phrases?) I increasingly hear in my circles is ‘agent-based workflows.’ Like a lot of buzzy projects, people can’t seem to fully agree what an agent-based workflow looks like exactly—other than it’s a thing that uses AI to get a complex-adjacent thing done. The logic for agents is straightforward: rather than having to engineer a series of prompts and other actions directly, just tell something to do that for you. Then said ‘something’ will design the prompts itself to build out a kind of checklist that it has to accomplish, go through those tasks one by one, and keep going until it completes a given task.

One way to look at the rise of AutoGPT and other agent-based AI technology is that these tools are merely super-charged versions of robotic process automation (RPA), a startup category that received prodigious funding over the past decade and led to decacorns like UiPath, which orbited 2021 at a $35 billion valuation before plunging back down to a more earthly valuation.

In parallel with my analysis about the future of the creative class in “Garrulous Guerrilla” (simply: most “creativity” is unoriginal and thus automatable), the vast majority of corporate work can be written out as checklists and simple logic. RPA platforms required consultants to investigate, design and implement procedures, but AI bots already seem capable of developing and running checklists themselves with little human intervention. The great corporate productivity hope for AI is that a series of tasks that formerly took busybodies several hours to do through email and enterprise systems of record could be effortlessly finished in seconds by an intelligent bot.

Let’s get a bit more speculative though. What if the efficiency here was less valuable for corporate productivity, and actually much more useful for labor organizing? In short, what if the next labor organizing came not from Josh’s specter of a Jimmy Hoffa-esque labor leader, but rather from a chatbot? Or as I dubbed it, HoffaGPT.

After all, labor organizing — like all political organizing — combines persuasion with a dedicated set of steps to turn intention into action, and even persuasion itself these days is often a battle-tested checklist. Organizing groups develop scripts, messages, narratives, and simulated role-plays in order to improve the persuasiveness of their activists in the field. Agent-based workflows are a ready match.

Now, HoffaGPT may not replace the emotional resonance of a labor leader inspiring legions of workers to overthrow their shackles and demand better dental coverage. But there’s no doubt in my mind that the same productivity enhancements that corporations are expecting from AI could absolutely improve the organizing potential of nascent unionization efforts.

Chatbots — right now — can functionally track people, schedule outreach, develop new slogans and offer better messaging scripts, answer questions both logistical and philosophical, as well as organize pickets and publish collective media responses. Combining tasks is also feasible as AutoGPT has shown, and so it’s merely a question of when — not if — the first agent-based workflow for unionization comes to fruition.

Even as workers attempt to ban AI, that technology could ironically buttress those very workers in their own fights for better salaries and working conditions. Far from exacerbating the gap between management and laborers, AI can and should empower workers to replace their managers entirely.

Frankly, it is middle management itself that will need to unionize in the face of an AI-driven productivity decimation.

This is a theme I have seen emerging for almost a decade now. Back in 2014 at TechCrunch, I wrote “Algorithms Are Replacing Unions As The Champions of Workers,” focused on the benefits that well-designed online labor marketplaces could offer workers in terms of job flexibility, a desired characteristic that has only become more prominent in the decade since in our post-Covid, remote-centric working world.

There’s almost universally negative coverage of freelancing and algorithmic marketplaces, but so much of that dissension evades what workers actually want. Yes, they need health and financial security (the very kind of interconnected securities this aptly-named newsletter likes to harp on), but beyond those requirements, most workers want flexibility and the ability to pursue different types of work. Unions, given their focus around centralized workplaces, have struggled to avoid shrinking as work has become more diffuse. AI is the structural shift that allows us to evolve from hierarchical, command-and-control corporations with rigid labor schedules to networked forms of work that can be stretched around our individual lives.

At least, that’s the dream. One thing is clear though: further labor strife is coming, perhaps most notably the looming deadline between UPS and the Teamsters in July, which could see hundreds of thousands of parcel carriers on the picket line and millions of online retailers struggling to get their products into the hands of consumers. HoffaGPT is already lurking, and whether labor or management ultimately is the winner remains very much to be seen.

Apocalypse NYC

The view from Lux's NYC office this week. Photo by Danny Crichton

The “Securities” newsletter is about unraveling the complex networks of risks that cascade to create singular moments of opportunity or crisis. There’s probably no better example of that focus recently or locally than the ash-strewn skies of Manhattan, which have been smothered in Canadian wildfire pollution the past week and plunged the city at times into mid-day darkness.

There’s nothing unique about these urban smoke-outs: the same pattern took hold over San Francisco in 2020 amidst the pandemic, and Singapore has been dealing with the same challenge from Indonesia’s wildfires for years now.

I have no deep insight here that hasn’t been repeated ad nauseam, other than to point out the sheer complexity of the problem in the “Securities” tradition. Global climate change has dried out Canada’s forests, transforming lush boreal carbon sinks into tinderboxes ready to spark into stunning conflagrations with a single lightning bolt. Protecting those boreal forests is — at this point — impossible, and humans are increasingly at the whim of stronger and less predictable winds carrying burning detritus.

The effects are myriad. Just on the little slip of island that is Manhattan, the ash halted flights at LaGuardia and Newark (delaying our own Deena Shakir, who is presenting at several NYC conferences), canceled outdoor sporting events, school recess and then in-person learning, as well as shut down restaurants, stores and public libraries. From personal observation, streets were relatively deserted, subways were emptier, and it seemed as if life had returned to moments during the Covid-19 pandemic. Worse, this social and economic crisis is outside the ability of federal and local leaders to ameliorate. Helplessness and powerlessness are pervasive feelings.

Now multiply this pattern across the entire urban northeast United States.

A new research paper in Nature last week emphasized that we have already crossed many of the barriers keeping the world safe in the years ahead. “Seven of eight globally quantified safe and just [Earth system boundaries (ESBs)] and at least two regional safe and just ESBs in over half of global land area are already exceeded,” the authors wrote. Expect more and worse, and an ever increasing need for investment in wide-scale societal resilience.

Bio x ML Hackathon

Lux was proud to back and help organize the Bio x ML Hackathon last week alongside our portfolio company Hugging Face and many others. Congratulations to all the winners and the many brilliant entries (read the project summaries available on the site). There’s electric momentum at the intersection of biology and machine learning these days, and the next generation of computer scientists and life scientists are quickly propelling the field forward.

Impossible to Inevitable Event in LA

Lux was delighted to host so many ambitious hard science and deep tech builders in Los Angeles this week as part of LA Tech Week, where the limelight is ever shifting toward aerospace and defense.

Lux Recommends

While sticking to the AI theme, our scientist-in-residence Sam Arbesman points to Isabelle Bousquette’s story in the WSJ’s CIO Journal about how generative AI will lead to a crisis of code maintenance. “’People have talked about technical debt for a long time, and now we have a brand new credit card here that is going to allow us to accumulate technical debt in ways we were never able to do before,’ said Armando Solar-Lezama, a professor at the Massachusetts Institute of Technology’s Computer Science & Artificial Intelligence Laboratory.”
The Nuclear Emergency Support Team is a unique advisory group at the Department of Energy that helps all federal agencies deal with the consequences of nuclear fallout. Two executives of the program talk about its inception and expansive role across man-made disasters and terrorism globally in “Nerds, Ninjas, and Neutrons.” “Although NEST is usually depicted in film as a shadowy unit complete with black helicopters and disguised personnel, the reality is at once more mundane and more impressive.”
Tess Van Stekelenburg recommends two articles from the Bio x AI world. She first highlights Geneformer, a new, pre-trained ML model that can “accelerate discovery of key network regulators and candidate therapeutic targets” according to scientists in Nature. It was a popular tool at Lux’s recent Bio x ML hackathon mentioned above. Second, she highlights Elliot Hershberg’s writeup of the Blockchain Meets Bio Summit, which is heady on ideas around decentralized finance and bio manufacturing.
Our summer associate Reha Mathur points to a brand-new article in Nature (our third citation this week!) on how computer scientists have used deep reinforcement learning to optimize some of the most fundamental algorithms used on computers, namely sorting and hashing. “AlphaDev discovered small sorting algorithms from scratch that outperformed previously known human benchmarks. These algorithms have been integrated into the LLVM standard C++ sort library.”
Finally, our summer associate Dev Gupta offers up a post by Michael Trott at Wolfram comparing the performance of OpenAI’s code interpreter against the Wolfram Plugin across 100 challenging mathematics problems. What’s interesting is the final lines: “In conclusion, the ability of ChatGPT to inspect results, make corrections, and enhance its performance is indeed a significant strength. While it is not infallible and can occasionally make errors, its error correction mechanisms and the capacity to refine code based on test cases contribute to its overall efficacy.”

That’s it, folks. Have questions, comments, or ideas? This newsletter is sent from my email, so you can just click reply.

Forcing China’s AI researchers to strive for chip efficiency will ultimately shave America’s lead

In incididunt ad qui nostrud sint ullamco. Irure sint deserunt Lorem id officia dolore non. Anim dolor minim sit dolor et sint aliquip qui est. Ex in tempor laborum laboris dolor laboris ullamco quis. Enim est cupidatat consequat est culpa consequat. Fugiat officia in ea ea laborum sunt Lorem. Anim laborum labore duis ipsum mollit nisi do exercitation. Magna in pariatur anim aute.

In incididunt ad qui nostrud sint ullamco. Irure sint deserunt Lorem id officia dolore non. Anim dolor minim sit dolor et sint aliquip qui est. Ex in tempor laborum laboris dolor laboris ullamco quis. Enim est cupidatat consequat est culpa consequat. Fugiat officia in ea ea laborum sunt Lorem. Anim laborum labore duis ipsum mollit nisi do exercitation. Magna in pariatur anim aute.

‍

Right now, pathbreaking AI foundation models follow an inverse Moore’s law (sometimes quipped “Eroom’s Law”). Each new generation is becoming more and more expensive to train as researchers exponentially increase the number of parameters used and overall model complexity. Sam Altman of OpenAI said that the cost of training GPT-4 was over $100 million, and some AI computational specialists believe that the first $1 billion model is currently or will shortly be developed.

As semiconductor chips rise in complexity, costs come down because transistors are packed more densely on silicon, cutting the cost per transistor during fabrication as well as lowering operational costs for energy and heat dissipation. That miracle of performance is the inverse with AI today. To increase the complexity (and therefore hopefully quality) of an AI model, researchers have attempted to pack in more and more parameters, each one of which demands more computation both for training and for usage. A 1 million parameter model can be trained for a few bucks and run on a $15 Raspberry Pi Zero 2 W, but Google’s PaLM with 540 billion parameters requires full-scale data centers to operate and is estimated to have cost millions of dollars to train.

Admittedly, simply having more parameters isn’t a magic recipe for better AI end performance. One recalls Steve Jobs’s marketing of the so-called “Megahertz Myth” to attempt to persuade the public that headline megahertz numbers weren't the right way to judge the performance of a personal computer. Performance in most fields is a complicated problem to judge, and just adding more inputs doesn't necessarily translate into a better output.

And indeed, there is an efficiency curve underway in AI outside of the leading-edge foundation models from OpenAI and Google. Researchers over the past two years have discovered better training techniques (as well as recipes to bundle these techniques together), developed best practices for spending on reinforcement learning from human feedback (RLHF), and curated better training data to improve model quality even while shaving parameter counts. Far from surpassing $1 billion, training new models that are equally performant might well cost only tens or hundreds of thousands of dollars.

This AI performance envelope between dollars invested and quality of model trained is a huge area of debate for the trajectory of the field (and was the most important theme to emanate from our AI Summit). And it’s absolutely vital to understand, since where the efficiency story ends up will determine the sustained market structure of the AI industry.

If foundation models cost billions of dollars to train, all the value and leverage of AI will accrue and centralize to the big tech companies like Microsoft (through OpenAI), Google and others who have the means and teams to lavish. But if the performance envelope reaches a significantly better dollar-to-quality ratio in the future, that means the whole field opens up to startups and novel experiments, while the leverage of the big tech companies would be much reduced.

The U.S. right now is parallelizing both approaches toward AI. Big tech is hurling billions of dollars on the field, while startups are exploring and developing more efficient models given their relatively meagre resources and limited access to Nvidia’s flagship chip, the H100. Talent — on balance — is heading as it typically does to big tech. Why work on efficiency when a big tech behemoth has money to burn on theoretical ideas emanating from university AI labs?

Turn to China now, and the story is in a very different place. The Biden administration has placed a series of export controls on advanced computation and semiconductor fabrication to China, including blocking the export of Nvidia’s H100 and A100 chips. Nvidia responded in March by offering tweaked versions that comply with U.S. export controls (with what it dubs the H800 chip), but the general policy remains in force: prevent China from acquiring the best chips that Earth produces.

Without access to the highest-performance chips, China is limited in the work it can do on the cutting-edge frontiers of AI development. Without more chips (and in the future, the next generations of GPUs), it won’t have the competitive compute power to push the AI field to its limits like American companies. That leaves China with the only other path available, which is to follow the parallel course for improving AI through efficiency.

For those looking to prevent the decline of American economic power, this is an alarming development. Model efficiency is what will ultimately allow foundation models to be preloaded onto our devices and open up the consumer market to cheap and rapid AI interactions. Whoever builds an advantage in model efficiency will open up a range of applications that remain impractical or too expensive for the most complex AI models.

Given U.S. export controls, China is now (by assumption, and yes, it’s a big assumption) putting its entire weight behind building the AI models it can, which are focused on efficiency. Which means that its resources are arrayed for building the platforms to capture end-user applications — the exact opposite goal of American policymakers. It’s a classic result: restricting access to technology forces engineers to be more creative in building their products, the exact intensified creativity that typically leads to the next great startup or scientific breakthrough.

If America was serious about slowing the growth of China’s still-nascent semiconductor market, it really should have taken a page from the Chinese industrial policy handbook and just dumped chips on the market, just as China has done for years from solar panel manufacturing to electronics. Cheaper chips, faster chips, chips so competitive that no domestic manufacturer — even under Beijing direction — could have effectively competed. Instead we are attempting to decouple from the second largest chips market in the world, turning a competitive field where America is the clear leader into a bountiful green field of opportunity for domestic national champions to usurp market share and profits.

There were of course other goals outside of economic growth for restricting China’s access to chips. America is deeply concerned about the country’s AI integration into its military, and it wants to slow the evolution of its autonomous weaponry and intelligence gathering. Export controls do that, but they are likely to come at an extremely exorbitant long-term cost: the loss of leadership in the most important technological development so far this decade. It’s not a trade off I would have built trade policy on.

The life and death of air conditioning

Across six years of working at TechCrunch, no article triggered an avalanche of readership or inbox vitriol quite like Air conditioning is one of the greatest inventions of the 20th Century. It’s also killing the 21st. It was an interview with Eric Dean Wilson, the author of After Cooling, about the complex feedback loops between global climate disruption and the increasing need for air conditioning to sustain life on Earth. The article was read by millions and millions of people, and hundreds of people wrote in with hot air about the importance of their cold air.

A few weeks ago, Bloomberg had a great piece about the rising toll of air conditioning under the similarly scary headline, “A Billion New Air Conditioners Will Save Lives But Cook the Planet.” The numbers, as always, are staggering:

Demand for air conditioners is surging in markets where both incomes and temperatures are rising, populous places like India, China, Indonesia and the Philippines. By one estimate, the world will add 1 billion ACs before the end of the decade. The market is projected to
before 2040. That’s good for measures of public health and economic productivity; it’s unquestionably bad for the climate, and a global agreement to phase out the most harmful coolants could keep the appliances out of reach of many of the people who need them most.

This is a classic feedback loop, where the increasing temperatures of the planet, particularly in South Asia, lead to increased demand for climate resilience tools like air conditioning and climate-adapted housing, leading to further climate change ad infinitum.

And there doesn’t appear to be a way out. Scientists have determined that there is a zone of comfort for humans where we can function at peak performance. Much of the industrialized world happens to sit in these temperate zones, which is one argument for their long-term economic success compared to the tropics. But as climate change intensifies, the world’s temperate zones are expected to narrow dramatically, and some countries like India may have next to no temperate zones left to protect its people. Air conditioning is a necessity (with some even calling it a human right), but greener coolant alternatives are still in their infancy. In other words, cooling innovation needs to heat up in a big way.

Lux on the Farm

Josh Wolfe speaking at Nvidia Auditorium at Stanford. Photo by Chris Gates.

Josh Wolfe gave a talk at Stanford this week as part of the school’s long-running Entrepreneurial Thought Leaders series, talking all things Lux, defense tech and scientific innovation. The
.

Lux Recommends

As Henry Kissinger turns 100, Grace Isford recommends “Henry Kissinger explains how to avoid world war three.” “In his view, the fate of humanity depends on whether America and China can get along. He believes the rapid progress of AI, in particular, leaves them only five-to-ten years to find a way.”
Our scientist-in-residence Sam Arbesman recommends Blindsight by Peter Watts, a first contact, hard science fiction novel that made quite a splash when it was published back in 2006.
Mohammed bin Rashid Al Maktoum, and just how far he has been willing to go to keep his daughter tranquilized and imprisoned. “When the yacht was located, off the Goa coast, Sheikh Mohammed spoke with the Indian Prime Minister, Narendra Modi, and agreed to extradite a Dubai-based arms dealer in exchange for his daughter’s capture. The Indian government deployed boats, helicopters, and a team of armed commandos to storm Nostromo and carry Latifa away.”
In the happier news department of science, Ina Deljkic recommends “’Quite science fiction’: Brain-spine interface system helps man with paralysis walk again.”
Sam recommends Ada Palmer’s article for Microsoft’s AI Anthology, “We are an information revolution species.” “If we pour a precious new elixir into a leaky cup and it leaks, we need to fix the cup, not fear the elixir.”
I love complex international security stories, and few areas are as complex or wild as the international trade in exotic animals. Tad Friend, who generally covers Silicon Valley for The New Yorker, has a great story about an NGO focused on infiltrating and exposing the networks that allow the trade to continue in “Earth League International Hunts the Hunters.” "At times, rhino horn has been worth more than gold—so South African rhinos are often killed with Czech-made rifles sold by Portuguese arms dealers to poachers from Mozambique, who send the horns by courier to Qatar or Vietnam, or have them bundled with elephant ivory in Maputo or Mombasa or Lagos or Luanda and delivered to China via Malaysia or Hong Kong.”
Finally, Sam recommends Brick Technology’s video of “Making A Billion-Year Lego Clock.”

That’s it, folks. Have questions, comments, or ideas? This newsletter is sent from my email, so you can just click reply.

‍

June 3, 2023

Export Cooling

<- Previous

continue

reading

June 17, 2023

Apex Prey

Next ->