š® Hinton & deep learning; misinformation; meritocracy; zeposeconds, platypuses & the Pope++ #295
Hi, IāmĀ Azeem Azhar. I conveneĀ Exponential ViewĀ to help us understand how our societies and political economy will change under the force of rapidly accelerating technologies.
š„ Join me next Friday for a live 30-minute discussion with Bloombergās Editor-in-Chief, John Micklethwait, whose latest book, co-authored with EV reader Adrian Wooldridge, looks into how the Covid-19 pandemic has exposed Westās weaknesses. Weāll stream live on Linkedin, YouTube and Twitter.
š Members can now read theĀ transcript of my conversation with DeepMindās CEO, Demis Hassabis. Full podcast isĀ available to all to listen here.
⨠Earlier this week, I briefly wrote about Chinaās chip opportunity, and hosted a membersā discussion about the future of citizenship.
A reminder: Iām hosting two discussions at this yearās World Economic ForumāsĀ Pioneers of Change Summit.Ā EVĀ members get a free pass for the whole event and two months of complimentary Digital Membership on the Forumās platform.Ā Members can access this offer here.Ā
The near future
š¤ āI do believe deep learning is going to be able to do everything, but I think there are going to have to be quite a few conceptual breakthroughs. The human brain has about 100 trillion parameters, or synapses, whereas what we now call a really big model, GPT-3 for example, has 175 billion. Itās a thousand times smaller than the brain,ā says Geoff Hinton, one of the pioneers of deep learning. In an interview with the brilliant Karen Hao, Hinton made the claim. Critics of this view include the CEO and founder of Robust.AI, Gary Marcus, with whom I spoke with a year ago.Ā My view: I do think that weāll be surprised by what weāll do with deep learning. The test will come when GPT-n has more parameters than the brain which may be only five-to-ten years away. Will a 100-trillion-parameter neural network outperform a 100-trillion-connection brain? I doubt it. Will it be an amazing piece of technology? Certainly. What about a 10-quadrillion-connection neural network? What will that be like? And what will we call the things it can do?
šÆ Why do we need to end the data economy, and how do we do it? Last week Carissa VĆ©liz joined me to answer these questions. Iāll leave you with one thought from Carissa and hope that youāll listen to the whole episode: āPeople might think itās very radical to call for the end of the data economy. But what is extreme is a business model that depends on the systematic violation of rights.ā
š Misinformation was rife during the US presidential election, particularly when it came to the already fraught issue of voting and ballots. A group of researchers took five hours and 120 people to fact check and trace the origin of a conspiracy theory called SharpieGate. See also, an exposĆ© of how YouTube channels make money off fake and misleading livestreams of election results. YouTube has confirmed some channels will still be allowed on the site, although the owners will be demonetised.Ā
š§ An interesting profile of the CEO of Volkswagen, Herbert Diess, on the electrification opportunity to which VW has committed more than ā¬30bn. āThe danger is that electric cars, which contain far fewer parts than combustion engine models, will be commoditised. It is a scenario Mr Diess hopes to combat by owning the valuable customer data generated by vehicles that are ever more automated and connected to the web.āĀ
š CCTV firm, Dahua, developed software to identify and track Uyghurs, the minority Muslim population in China who have been subject to torture and violence from the state. The software source code was even publicly available on Github, although it now appears to have been taken down. Itās not even the first company to do so:Ā Hikvision, the number one CCTV manufacturer in the world according to Charles Rollet, did so last year.Ā
Dept of politics
Joe Biden won the US election but I was very curious about the forces that underly some of our politics, especially in the last decades of accelerating change. Here are some interesting views.
Philosopher Michael Sandel talks to The Guardian about where he thinks the left has gone wrong āĀ its pursuit and belief in meritocracy āĀ ahead of his new book, The Tyranny of Merit. The interview is fascinating. Sandel says:
On globalisation these parties said the choice was no longer between left and right, but between āopenā and āclosedā. Open meant free flow of capital, goods and people across borders. To object in any way to that was to be closed-minded, prejudiced and hostile to cosmopolitan identities.
See also Robin Hansonās dissection of prestige in a partial review of a new book written by Lauren A. Rivera.Ā
Fintan OāToole writes about what happens to democracy after the fact ā āin this frame of mind, there can never be a result of the 2020 electionā - about how Trump and his cronies have corrupted anything that comes next by creating āa great prairie of paranoia.āĀ
Voting patterns and presidential elections in the US have been impacted by a 100 million-year-old coastline. The characteristics of a particularly rich soil led to fertile ground for cotton farming, which was where most slaves were forced to work. In turn, that has led to greater numbers of African American people situated around where this coastline used to be, an area now known as the Black Belt.
Dept of quantum computing
Quantum computers boil down to qubits, the ābitsā that encode information and enable quantum algorithms to run various calculations at the same time, at an exponential scale. D-waves quantum annealing is being used by a grocery chain in Canada. This survey on progress in quantum computing is a helpful update to progress in the domain. The field is certainly widening, Honeywell, best known for home thermostats, has even launched a quantum computing project which has quite some credibility.
Peter Shor, one of the luminaries of quantum computing, spoke to Nature about his breakthrough in 1995, and warned that we are not prepared for a post-quantum world.Ā
š Dept of decarbonisation: 411.56ppm | 3,491 days
Each week, Iām going to remind you of the CO2 levels in the atmosphere and the number of days until we reach the 450ppm threshold.
The latest measurement (as of November 2): 411.56 ppm; November 4, 2019: 409.42 ppm; 25 years ago: 360 ppm; 250 years ago, est: 250 ppm. Share this reminder with your community by forwarding this email or tweeting this.
Iberdrola, the worldās third-largest utility company, has promised to double its renewable energy capacity over the next five years. It will invest around ā¬10 billion in 2020 and a further ā¬65 billion will be spent on āgrowthā. The chairman of Iberdrola has said that the renewable capacity will rise to 60 gigawatts in 2025.Ā
Platypus of the week
Some platypuses fluoresce when exposed to ultraviolet light.
Chart of the week will return next week.Ā
Short morsels to appear smart while counting votes
𤳠78% of people surveyed said that are interested in buying one of the new iPhone models.
China has already moved onto post-Covid fashion and seems to be embracing a cyberpunk look, with technologically influenced, high-glamour fashion for all genders. See also from China: an underreported side of Chinaās technological revolution is the relationship betweenĀ pearl farmers and hackers in rural China.Ā
𤩠See yaā later femtosecond braggarts. An experiment has measured zeptosecond-scale processes.
š A story from September, which suggested that Venusā atmosphere potentially had phosphine, a sign of life, may not be true ā the algorithm used to filter out noise in telescopes may have overfitted.
Do intellectual property laws imagine creatorship as the domain of whiteness? A new book argues that this is so.Ā
šš» Pope Francis has been praying for benevolent AI.
š Every increase of 1mg of fine-particle air pollutants in a US county was associated with an 11% increase in CV-19 mortality rate.
Whatās so good about being a polymath?
End note
Crazy week. Looking forward to another four years before I have to visit Drudge Report!
Stay well,
Azeem
What youāre up toānotes from EV readers
Greg Williams launches Wired World 2021 this week (I have a piece in it, and EV researcher Sanjana Varghese has stories in it too, as do many other EV readers. Itās quite an amazing piece of commissioning, recommended.)Ā
EV researcher Sanjana Varghese wrote about Feels Good Man āĀ a documentary on Pepe the Frog ā for The EconomistĀ and went on the daily podcast to talk about it.Ā
Mete Varas has written a short piece for Medium about what healthy food and fake news may have in common.
Rodolfo Rosini has created a DIY kit for smart masks. Upvote on Product Hunt!
Lucia Komljen writes explores how AI is changing the creation and consumption of culture.
Christian Liensberger is developing Trove, which can help developers source better data for artificial intelligence.Ā
Sue Wheat of climate solutions charity Ashden has launched Letās Go Zero, a campaign to help schools become zero carbon by 2030.Ā
Jason Parry has written a play for human and AI spectators.
Andrew Green of the GreenTech Alliance is hosting an event on November 11th on the Allianceās achievements since its launch six months ago and its plans for the future.Ā
Knut Wimberger has recently written about pandemics on his blog.
Obi Felten has been working on Amber, a project at GoogleX to better measure and monitor mental health.Ā
Email marija@exponentialview.co to share your news and updates with us.