🔮 The explosion in companies; predicting technology success; moral machines; beyond hard skills; Indian unicorns, billionaires & new words++ #189
Moral machines, talent
|Oct 28, 2018||Public post|| 1|
Dept of podcasts 🎧
Silicon Valley is this cocky teenager. [...] As we grow from a teenager to adult, we need to say: "We have a lot of power and influence and we need to be responsible stewards of that power and influence; not just retrospectively but also proactively."
LinkedIn founder and investor, Reid Hoffman, and I discuss how Silicon Valley companies grow and what they need to learn as they mature into dominant companies that clash with social and political institutions.
Dept of the near future
💥 The coming explosion in companies. Demand is growing rapidly and the supply is expanding. Mind expanding essay. We will likely see "a massive increase in the speed, number and complexity of companies, all powered by automation of company formation, and driven by demand from new technologies. As with the shift from paper to electronic stock trading, the automation of companies will have profound and wide-reaching effects for society as a whole."
🥇 Which technologies will succeed? Brilliant analysis by Rodney Brooks.
🚎 Moral machines: MIT's long-running 'trolley problem' simulator shows that moral choices are culturally specific. The largest study on machine ethics - more than 2.3m people from 233 countries participated - identified key cultural differences about whether to spare the young, the old, pedestrians, property, the law-abiding or the unfit. Cultural difference around these choices were likely to emerge, but it is useful to see some large-scale data. While it raises practical questions, I think it's also important to recognise that automated systems (such as driver assistance) could reduce the incidence of any time of harm, even before we get to the trickier trade-offs of a trolley problem. It's also relevant to recognise the significant lack of consensus of what "ought to happen" in those trade-offs. Closing that latter loop will require new tools from designers and better societal articulation of how to understand those trade-offs.
⚡ Apple's Tim Cook unloads on business models relying on user data, warning that "[o]ur own information [...] is being weaponized against us with military efficiency.” (I touched on Apple's privacy advantage back in June 2016 in EV#67) Cook also calls for US privacy laws to better encompass the protections of Europe's GDPR, something that Microsoft and, even, Facebook has hinted at. Two things Cook doesn't mention which are relevant: the first is that personal data aggregated could form a collective good, so while we need mechanisms to prevent this data collection to be used against us, we could also benefit from mechanisms that allow us to tap into the positive externalities that may emerge from aggregate insights. One good example being health data. The second is that strong as European privacy law is, it may not protect us from what researcher, Sandra Wachter, calls 'privacy-invasive and unverifiable inferences'.)
🇺🇸🇨🇳 To escape the AI cold war, the US and China must embrace each other as partners, urge Nicholas Thompson and Ian Bremmer. "It’s not hard to see the appeal for much of the world of hitching their future to China. [...] China’s plans for a tech-driven, privacy-invading social credit system may sound dystopian to Western ears, but it hasn’t raised much protest there. [...] 84 percent of Chinese respondents said they had trust in their government. In the US, only a third of people felt that way." (Americans on both sides of the aisle trust Amazon more than the government according to this survey.)
🌐 A really relevant question as we grapple with many supranational issues and areas of national competition is whether we should "give up on global governance?" Jean Pisani-Ferry of the Bruegel Group suggests the answer is not quite but we do need to find the narrow zone of agreement that provides the "minimum basis for collective action." (Longer paper here is worth mulling over.)
⚖️ Marianna Mazzucato: We need a new way to measure value (See also: we need a new way to vote! On the Australian case for moving past democracy to sortition/citizen-juries.)
Dept of pay-it-forward
Exponential View grows solely from your personal recommendations. Helping EV grow helps EV give you more.
Do this by sharing via Twitter, LinkedIn or Facebook with a recommendation. Or by forwarding this email to several friends.
We're approaching 35,000 readers. I'd like to get to 50,000 by the end of the year.
I can only do that with the help of each and everyone of you reading these words. So, if you are reading them, please take a moment to make a recommendation. 🙏
Dept of talent
Jeff Wiener, the boss of LinkedIn, says that computers are not going to be taking all our jobs quite yet. “We’re still a ways away from computers being able to replicate and replace human interaction and human touch.” The result? Most in-demand, according to Wiener, are the so-called “soft skills”. (As an aside, while many prognosticators, myself included, think those skills of interaction and humanity are vital, recent US data still shows that most in-demand roles are emphasising highly technical, non-human-interaction skills like engineering, IT and trucking.)
One way to understand Jeff's comments is to recognise we actually live in a world populated by people. We need “soft skills”, like communication and leadership. But the term "hard skills", especially in opposition to "soft skills", is truly misleading. Hard in this context has two connotations. The first is difficult (as in "understanding the generalization of the chain rule to tensors is hard"). The second is that you're dealing with things like machines, that are physically hard ("I hurt my head when I hit it on the laser printer which is hard"). The trouble is that these terms put "soft skills" into an oppositional position. If hard means difficult (like tensors), then soft must mean easy.
Yes, soft skills involve working with things that are physically soft relative to server racks and diesel generators, like your boss or your team. The undercurrent that "soft means not hard means easy" remains, especially as we glorify the hard skills of STEM. In a world of well-defined objectives and militaristic commercial language, “soft” skills are harder to quantify and fit in a spreadsheet, and so have no place in the modern, Taylorist workplace.
Here's the thing. Look at what is in the bucket of soft skills: team management, interpersonal communication, empathy, conflict resolution, critical thinking, perspective-taking. The so-called “soft skills” are neither easy nor are they out of place in an organisation. The origin of the dichotomy comes from a US Army assessment between 1960-1970. Hard skills were hard because they were well-defined and straightforward. Soft skills were soft because “we don't know much about the physical and social environments in which the skill occurs and...the consequences of different ways of accomplishing the job function”.
In fact, the hard skills are actually the easy ones to grasp. You can wrap your head around them. Partial differentiation, or understanding the chain rule, or fiddling with a Gantt chart is really teachable and manageable once you know how. What is truly tough is persuading a child to do something they don't want to do. Or resolving a conflict between two, three or more people. Or motivating a recalcitrant team to follow you, even when the data doesn't support it. Or deciding that even when the data suggests it, something shouldn't be done. That's hard.
Hard skills are easy because they can be taught more easily, and often in scalable ways. So-called soft skills are tougher to explain and require considerably more complex modalities than "hard skills".
I'd propose a rebrand. “Soft” is the wrong modifier. These not-hard skills are actually the skills that we most need as individuals, at home and in the workplace. I'd go for "power skills". It encapsulates both what these skills give you and what it takes to master them.
Demand for blockchain engineers rises 400%. But as EV reader, Kevin Werbach, whose new book on blockchain is out next month, says, “No one is using blockchain, it doesn’t solve any problems, and crypto prices are down, yet somehow demand for developers is off the charts."
Demand for python coders amongst autonomous vehicle firms is so hot right now. (Demand for AV engineers is up 600% in the US in the past three years.)
Do you know how much your colleagues earn? Fair compensation starts with greater transparency, says Jackie Luo, as she shares as much Silicon Valley salary data as she can get her hands on.
Dept of robots
A comprehensive report on robotisation from the International Federation of Robotics merits a small department on its own. Industrial robotisation is growing, and service robots (think autonomous trolleys that can whizz around factories and hospitals) are a hot area.
Three key short-term technical developments to watch are (a) smart grippers (b) better connectivity (c) easier programming, in particular, by demonstration.
381k industrial robots shipped in 2017 at an average price of $43k for a total value of $16bn. +14% forecast CAGR to 630k industrial by 2021 (73% in APAC), est. 3.8m robots in factories worldwide by then;
China is the dominant supplier with 137.9k industrial robots in 2017, growing at 59% YonY;
Primary drivers are the automotive and electronics industries;
Highest robot density in South Korea (710 robots / 10k humans); world average of 85. US is at 200, China 97, the UK at 71. (Important to understand how this reflects the relative scale of the manufacturing base where physical robots are currently useful);
Professional service robots, in logistics, agriculture and health, are a fast-growing sector and by shipments should exceed industrial robots by 2021 (736k vs 630k). By value, service robots will represent about three-quarters of all robotics shipments;
In 2017, medical robots are the most valuable, clocking in at $660k per unit compared to an average of $55k for other service robots & $43k for industrial robots. Exoskeletons, as a category, are one of the fastest growing subsectors.
Short morsels to appear smart at dinner parties
Tesla finally turns a profit. "Like a magic trick, [Elon Musk] is quietly smoking the whole automotive industry." Worth reading this analyst report.
🔆 "The benefits of screens as a learning tool are overblown, and the risks for addiction and stunting development seem high." Silicon Valley insiders turn against screens for the kids.
🦄 There is a boom creating new tech unicorns in India—many are focusing on local issues.
Apple has paid $100bn to developers, and its service business is growing faster than its own 2016 forecasts. Oh, and it makes $1bn a month of pure profit from Google alone.
Detailed, technical critique stomps on Bloomberg's story about the Chinese chip hack.
One way to track the rise of tech giants is through the rise of corporate expense reports. (UberEats is spreading like a weed in the US.)
💸 The wealth of the world's billionaires grew 20% last year.
The US Air Force has been admonished for buying coffee cups for $326,000.
💾 Playlist, wafflestomper, woke, floppy disk, overhype. What words were first used the year you were born?
I had a few days off this week as it was half-term with my kids. Exponential View is a little shorter than usual.
If you have time and haven't already, I strongly recommend dipping into the current podcast season. There are some brilliant conversations you can enjoy. They work particularly well on the commute or, indeed, just while you are wandering around the house.
Have a great week,
P.S. I know we've been a little light on AI content the past few weeks. I've been thinking hard about a couple of other things.
P.P.S. Thanks for paying attention and taking a moment to share EV with your friends. If you have an extra minute, scroll down to learn about your fellow readers' projects: our community of readers it quite inspiring! Share your work by emailing Marija.
What you are up to
Sunny Hundal's thought-provoking piece on why the existing media narratives around climate change are counter-productive.
Several EV readers contributed to The State of the Token Market report.
Roger Nolan is looking for the VP of Data Science at Culture Trip, one of the fastest-growing companies to watch according to Forbes.
Jan Erik Solem's Mapillary is helping urban planners and policy makers increase green areas in the UK.
Codrina Cretu, researcher at Nesta, wrote to us about Nesta's new analysis of digital social innovation landscape in Europe.
Charles Thomas and Benjamin Perrin shared their vision for the future of work in this post-freelancing manifesto.