🚀 This is issue #104 of THE EXPONENTIAL VIEW. Sign-up for the newsletter here.
Darknet is more decentralised than Internet
🙊 Benedict Evans: Voice & the uncanny valley of AI. “People want voice to be the new thing …. the tech industry is casting around looking for the Next Big Thing. I suspect that voice is certainly a big thing, but we’ll have to wait a bit longer for the next platform shift.”
👨🏽🌾 How superstar firms have shrunk labour’s share of the pie. “These superstars can offer more variety, cheaper prices and convenience, but the bigger chunk of profits that they capture is split among fewer workers.”
👫 Will robots help our hinder the creation and maintenance of meaningful human friendships? John Danaher takes an Aristotelian view.
🌟 Tim Berners-Lee: We need to take control of our data, tackle misinformation and manage political messaging in order to save the Web. (Paradoxically, the Darknet now has a more decentralized structure than the mainstream Internet which may make it more resilient to attack or oligopoly power.)
🔥 Facebook wants more power in our lives, and we should resist, argues Jan Dawson. “It finally recognizes its power and the ways in which that power has caused problems in the world, but its instinct is to wield that power even more, rather than back off.”
Dept of artificial intelligence
Interview with Andrew Ng from Baidu and Neil Jacobstein from Singularity. Says Ng:
Just as about 100 years ago electrification changed every single major industry, I think we’re in the phase where AI will change pretty much every major industry.
And he continues
Things may change in the future, but one rule of thumb today is that almost anything that a typical person can do with less than one second of mental thought we can either now or in the very near future automate with AI.This is a far cry from all work. But there are a lot of jobs that can be accomplished by stringing together many one-second tasks.
📈 Put this together with Michael Schrage’s argument in Harvard Business Review that “machine learning and AI algorithmic innovation transform analytics … to supercharge the [Pareto 80/20 principle]”, and it paints a picture of how important it is for firms to get the grips with AI’s impact.
In essence, the availability of data across organisations will identify opportunities where the optimisation is not an 80/20 but perhaps a super-pareto (a 99/1). In other words, taking Pareto’s observation that small components may have substantial impact and lasering in on where those effects are more significant.
🔮 To see how this works in practice, I really recommend Danny Lange’s excellent presentation on AI in the enterprise. Lange has run machine learning platforms at Microsoft, Amazon, Uber and Unity. (I recently saw him speak at an excellent Nordic.AI event.)
The arrival of AI is also changing organisational structures in firms.
Many companies are now putting greater emphasis on cultural fit and adaptability, knowing that individual roles will have to evolve along with the implementation of AI.
Both Ng and Schrage describe AI in the terms that Jerry Kaplan calls “potent tools that promise a more prosperous and comfortable future”. Kaplan reckons that “[p]ublic discourse about AI has become untethered from reality.”
Or as Ines Montani says in this lovely essay: “please, finally, stop using that same old fucking wired brain illustration.”
Elsewhere:
- 🚗 How Drive.AI is using deep-learning from soup-to-nuts in their new autonomous vehicle tech. There has been much discussion about whether deep learning on its own can handle the complexities of autonomous driving or needs to be married with more traditional robotics approaches. Drive seems to be one firm emphasising a holistic deep learning approach. FASCINATING
- 🗼 How could artificial intelligence change architecture? AI could improve the understanding of a site and its environs, support design decision making and the creative endeavour, foster better client relations and facilitate novel AI-robot enhanced craftsmanship.
- Flippy the burger making robot. Moving into trial in 50 burger joints over the next two years. (Video)
- Nice application of machine learning to assist in fairer bail and sentencing
- IBM’s continuous speech recognition system achieves a 5.5% error rate
- A look at Graphcore’s chip architectures which are optimised for machine learning
- Did Cambridge Analytica’s secret psychographic sauce really help Trump’s victory? Possibly not.
- 48m twitter accounts are actually bots
- Shanghai factory plans to replace all humans with bots
Quantum dept of maybe
I’ve been meaning to cover more quantum computing for a few weeks but this week Nature and my buddies at The Economist went to town on it. You can access Jason Palmer’s survey in The Economist here.
- IBM has announced IBM Q, which promises to make quantum computing more widely accessible. It won’t outperform conventional computing (yet) but could be a useful playground for developers.
- ArsTechnica has more details on IBM’s 50-qubit computer.
- Time crystals, the seventh state of matter, are official. (We hinted at this in EV#98). Simple explanation from WSJ; more detailed explanation from Nature.
- 🔬 Obituary of Peter Mansfield, pioneer of magnetic resonance imaging
Small morsels to appear smart at dinner parties
We are watching more video, just not more TV
The Wintel marriage is falling apart and Microsoft is turning to new chip architectures at Intel’s cost.
Handy brief history of Blockchain
Tesla reckons it can solve South Australia’s energy problems within 100 days.
🇨🇳 China will have more than a quarter of a billion over-60s within 3 years
How blockchain might improve the food (and wider) trade
Your name might affect your facial appearance. (Machine learning meets psychology with a surprising result.)
Is ambiguity a better barometer of market uncertainty than volatility?
UK carbon emissions drop 6% as coal use tails off
Half of all new Norwegian cars are electric or hybrid
🌍 Brian Whittaker: The rise of Arab atheism
Tinder Select is a new invite-only version for highly-rated members.
👨🏽🔬 Gorgeous video of a neuroscientist explain the connectome to five different people (including a five year old kid).