At a Rise of AI Virtual Chat recorded live on 22 May 2020, host Fabian Westerheide talks with science-fiction author and technologist Daniel Jeffries about life after the first wave of COVID-19. Jeffries argues the pandemic is not past its peak and predicts further supply-chain disruptions, business closures and mass unemployment, especially in countries without strong health infrastructure. He describes a results-only work environment as the ideal setup for knowledge workers, criticizes commuting and most in-person business conferences as low-value, and says virtual events have cut production costs dramatically. He warns that crisis-driven surveillance and contact-tracing technology, once introduced, tends to persist through open-ended "sunset provisions," citing the USA PATRIOT Act as precedent, and criticizes governments and health authorities for downplaying the effectiveness of masks. On the upside, he expects accelerated open science, faster drug and antibody development, and overdue digitalization of bureaucracy. Asked whether decentralized AI could prevent misuse, he answers no, because algorithms simply encode the biases of the people who build them, regardless of how centralized the system is. The conversation closes with a question on the tyranny of the commons applied to mask-wearing.
Key takeaways
- Jeffries predicts a wave of business and restaurant closures as venues reopen at reduced capacity with already thin margins.
- He argues crisis-era surveillance and contact-tracing powers tend to persist indefinitely, citing the USA PATRIOT Act's open-ended sunset provisions.
- He calls a results-only work environment, not fixed office hours, the ideal setup for knowledge workers.
- He says decentralizing AI training is possible, but decentralizing governance would not make AI fairer, since algorithms encode human bias regardless.
- He expects immunotherapy and antibody cloning, long funded by DARPA, to accelerate faster than traditional vaccine development because of the pandemic.
Detailed notes
Entering lockdown
Fabian Westerheide recalls that the Wuhan lockdown of 30 million people in January 2020 was the moment he realized something serious was happening, prompting him to start stockpiling supplies. Daniel Jeffries says he initially thought the outbreak was overblown, along with most people outside epidemiology, and does not mind having been wrong. Westerheide describes the sudden closure of borders, schools, courts and businesses across Europe in March, and the loss of basic civil rights such as freedom of assembly and movement, comparing the scale of disruption to wartime.
What is lost: social life, travel and offices
Jeffries, who was living a location-independent lifestyle out of Airbnbs in Berlin before the pandemic, says he misses socializing in cafes and clubs and the ability to travel freely more than anything else. He argues that globalization, illustrated by a smartphone's components coming from 40 to 100 countries, cannot simply be replaced by localized production, even though it has real downsides. On remote work, he says a results-only work environment is the ideal setup for knowledge workers, and calls daily office commuting for people who do not build things with their hands an outdated habit. Westerheide adds that some of his portfolio companies have moved to fully optional office attendance with hot-desking, replacing daily commuting with occasional social events.
Supply chains, closures and unemployment
Jeffries reports growing meat and grocery shortages in the US, including fast-food chains unable to serve their core menu items, and expects further disruption to complex products sourced from many countries. He points to explosive case growth in Russia, Pakistan and India, and argues third-world countries lack the infrastructure to manage both the virus and the resulting supply shocks. He predicts a wave of business closures as restaurants and shops reopen at reduced capacity with thin margins, and cites 36 million people unemployed in the US within weeks. Westerheide contrasts this with Germany's short-time work scheme, which pays up to 80 percent of salary to discourage layoffs.
Information warfare and surveillance
Jeffries argues that societies with a culture of "information warfare" struggle most with the pandemic because false narratives compete with reality, and cites the CDC's own pandemic playbook as instructing officials to downplay mask effectiveness to conserve supply, a message he calls a lie. He criticizes the anti-mask movement in strong terms and calls for governments to mass-produce N95 masks domestically. He warns that pandemic-driven surveillance and contact-tracing programs, once passed, tend to remain permanently through what he calls sunset provisions that never actually sunset, using the post-9/11 USA PATRIOT Act as an example of emergency powers passed almost unanimously and never repealed. He also references a claim that Hong Kong protests were suppressed under cover of the outbreak.
Opportunities: virtual events, science and automation
Jeffries expects an acceleration of open science, noting that the virus's genetic sequence was shared immediately and amateur scientists worldwide could analyze the data. He anticipates faster progress in immunotherapy and antibody cloning, an approach he says DARPA has funded for years, as an alternative to traditional vaccine development. Westerheide notes that corporate clients are shifting large event budgets, for example cutting a two-million-dollar physical event budget to 1.5 million for a virtual version, since production costs fall by around 99 percent without venue, catering or crew. Jeffries expects a rise of hybrid formats combining virtual sessions with small local gatherings, alongside continued automation in meatpacking and other physically dangerous jobs, and growth in vertical/indoor farming and behavioral-analysis software.
Q&A highlights
Asked whether decentralized AI could prevent misuse, Jeffries says no: decentralized training is feasible, for example crowdsourcing labeled clips to teach a system to walk, but decentralizing governance does not make AI fairer, because every algorithm encodes the biases of the humans who build it, and every AI application is inherently dual-use. Asked about mask-wearing, Jeffries invokes what he calls the tyranny of the commons: individual decisions not to wear a mask carry little personal penalty but degrade collective protection for everyone else, the same dynamic he compares to overuse of antibiotics.
Quotes
I think a results only work environment is the ideal work environment for knowledge workers.
Daniel Jeffries
Reality is the greatest video game of all time.
Daniel Jeffries
It's called the tyranny of the Commons.
Daniel Jeffries
Questions answered in this talk
Can decentralized AI prevent AI from being misused?
According to Daniel Jeffries at this Rise of AI Virtual Chat, no: decentralizing the training of AI systems is technically possible, but decentralizing governance does not make AI fairer, because every algorithm reflects the biases of the humans who build it, and AI itself is a dual-use technology.
Why does Daniel Jeffries think pandemic surveillance measures will not go away?
Jeffries argues at this talk that governments rarely relinquish emergency powers once granted, pointing to the USA PATRIOT Act as an example of a law passed almost unanimously after 9/11 under a nominal sunset clause that in practice never expired.
What is the ideal work model for knowledge workers according to Daniel Jeffries?
Jeffries states in this conversation that a results-only work environment, judging people by what they deliver rather than fixed office hours, is the ideal setup for knowledge workers, and calls mandatory commuting to an office outdated for people who are not working with their hands.