🔮 Becoming a centaur; creative destruction; virtual citizens; why decentralisation matters; power laws, peak oil & passwords ++ #154
|Azeem Azhar||Feb 25, 2018|
Dept of the near future
🎠 How to be a centaur: Intelligence augmentation will allow “for new, previously-impossible ways of thinking, of living, of being”. One lovely insight: “AIs are best at choosing answers. Humans are best at choosing questions.” MUST READ
🔥 Why decentralization matters. Chris Dixon argues that decentralized cryptonetworks can be fundamentally aligned to user welfare in ways that centralised platforms can’t be. “[T]he hardware-based networks of the past are fundamentally different than the internet, a software-based network. Once hardware-based networks are built, they are nearly impossible to rearchitect. Software-based networks can be rearchitected through entrepreneurial innovation and market forces.”
🛂 The Atlantic on buying and selling virtual citizenship. (See also: Digital nomads are “firing their governments” and moving to climes that better suit their purposes. A luxury - to choose citizenship - that had previously only been enjoyed by the ultra-rich is becoming marginally more widely accessible.)
🏚️ The corporate longevity collapse continues. The average tenure of a firm in the S&P500 was 24 years in 2016 and is forecast to halve by 2027.
🌥️ Stablecoins are a way of addressing the volatility experienced by most cryptocurrencies. Here is a useful introduction to the three mechanisms we might implement some kind of price-stability-by-design in crypto.
Dept of AI and automation
Automation is transforming the garment industry in emerging economies, savaging local worker numbers and hinting at a reshoring trend. (Excellent reportage.)
🔬 Deep learning in biology, with all its perks and flaws. “Another challenge with deep learning is that the computers are both unintelligent and lazy, [...] they lack the judgement to distinguish biologically relevant differences from normal variation.” Super solid survey.
The Malicious AI report outlines several key areas of real risk from AI systems, including increased affordability of cyber attacks, controlling malicious hardware at scale (think drone swarms) and new methods for spreading disinformation. See full report here.
Visualising deep learning models at Facebook. A really interesting tool which will help debug and explain the operations of a deep neural net.
Reinforcement learning is a really interesting area of study. Alex Irpan draws attention to its current limitations: data inefficient; the challenge of specifying & designing a reward function for many activities; escaping local maxima; risks of overfitting & reproducibility.
There are only 22,064 PhD-trained AI experts in the world, pointing to huge upcoming shortages of the talent.
💯 Dynamic word embeddings for evolving semantic discovery. Lovely overview of a recent paper.
Short morsels to appear smart at dinner parties
Do power laws, which represent highly unequal distributions in networks, exist in real-world networks? Some evidence not. And evidence that you can design networks to mitigate power law effects if you want.
🤳 Selfies and wefies: why Chinese handsets are putting the iPhone on the back foot in Asia. In other Apple news, the company raises human rights watchers' fears as it begins storing iCloud keys in China.
The Dunedin study, one of the largest longitudinal studies of human health, is giving up some secrets of longevity & quality of life. (Self-control is a key one.)
💊 Anti-depressants do work, says a major metastudy.
As tech wrings its hands in a “what have we wrought moment”, Audrey Watters savages the growing tech regrets confessional.
BP: peak oil in 2040: 15% of vehicle miles will be electric & about 25% of journeys autonomous. (Encouraging, but I think they have been too conservative about the energy switch.)
🤣 Words of Heart is a dating site which matches you to people who use the same password as you.
I had a discussion earlier this week with some luminaries involved in science communication. I shared one of my concerns about the current backlash against Technology: the guilt by association, the rejection of science and small-technology. The handy epistemic shortcuts which, in turn, favour talking heads and media sound bites, that connects the hubris of, in particular, Facebook to the very idea of scientific progress (which is still pretty darn cool, and certainly very pretty.)
We (all of us) have got a lot of work to do to rebuild faith and confidence in the process of science, the gains from research, the value of the entrepreneur and the power of technology but perhaps through the lens of some reasoned and challenging criticality.
Separately, EV has had some problems with rogue unsubscriptions from the newsletter last week when people forwarded it to their friends. We’re trying a new method to forward it to friends. Could you please give it a shot by clicking here.
Have a great week,
P.S. We still appreciate your tweeting about the issue :)