🔮 Losing the information war; business in a box; CRISPR tinkering; principles of an AI society; fasting, vegan meat & decisions++ #194

Drawing a digital Maginot line

Dec 2, 2018Public post
This issue has been supported by Hanover Communications, protecting your brand in the age of AI.

Dept of the near future

⛔ The digital Maginot line. Renée DiResta argues that "the Information World War has already been going on for several years... the combatants are professional, state-employed cyber warriors and seasoned amateur guerrillas pursuing very well define objectives... destabilising societies in some countries." Further, she argues "what made democracies strong in the past - a strong commitment to free speech and the free exchange of ideas - makes them profoundly vulnerable in the era of democratised propaganda." We don't, yet, have clear solutions but "low-level roadblocks". If you read one thing this weekend, read this. (Azeem's comment: Brave thinking in this essay. And not too hyperbolic. Platforms can obviously do more to tamper disinformation. But more deep-seated inoculations are critical thinking skills, as part of a toolkit of deep media literacy, & high trust levels in society. They have a very long lead time to develop. More solutions will be needed, especially as Sara Fischer argues "misinformation bots are smarter than we thought")

📊 Public data for public good: Barcelona launches a pilot program, led by CTO Francesca Bria, which means transit companies, like bike-sharing firms, are being required to share their data to be published on an open data portal as part of their contracts with Barcelona city authorities. The goal is to turn privately collected data into a resource for public good, as well as give increased control to citizens over the use of their data. This is such an interesting experiment in building a digital commons. Prediction: many more of these to come around the world. (Request: if you are building a city- or national-level data commons, or other type of data co-operative or commercial data sharing platform, I would like to know.)

🔄 How marketplaces might reinvent the $10trn service economy. Online marketplaces have evolved from simple classified listings (of the Craigslist era). We've seen specialisation (e.g. Airbnb) and marketplaces taking on more parts of the value chain (trust and verification, real-time matching, identity, payments and even, in the case of firms like Opendoor, assuming risk & market-making). The next phase, argues Andrew Chen, will be to enter the service economy which in the US alone is 125m jobs. Automation & AI could play a key role here as the marketplace might use those technologies to fulfil simple needs automatically. (For a prescient argument for AI-based service marketplaces, look at Fabrice Grinda's 2016 article, The Next Evolution.)

💈 The "business-in-a-box" could be the next stage in franchise businesses, argues investor Saar Gur. These "platforms give entrepreneurs the opportunity and infrastructure to run a business without the convoluted and expensive process of building a company from scratch". Similar to franchising, but it includes more than branding and standard-operating-procedures: infrastructure, back-office and software tools.

🏢 AI thinks like a corporate, and that's worrying, says Johnnie Penn. A thought-provoking argument on the "noticed uncanny similarity between the functionality of a human bureaucracy and that of the digital electronic computer." (Azeem's comment, as I've argued in EV previously, if we want to understand how an AI system with a simple objective function will behave, with regard to unintended consequence, a good analogy is that of corporations managed solely with a focus on the single-objective function of profit-maximisation, and without regard to other stakeholders or the environment.)

🤑 At a $100,000 per Instagram post, influencers can make real money. Why? The very high engagement rates, sometimes reaching 30 percent. (Influencers aren't infallible: discount shoe store, Payless, pranked LA-influencers with an Italian-sounding shop called Palessi. The result? Self-styled bastions of trend paying $500 for twenty-buck clogs.)


Dept of appreciation 👐

The EV newsletter is provided for free. If you are reading it and value it, please make sure that you have shared it to your social networks (Facebook, LinkedIn or Twitter) or by email to friends at least once in the previous four months. Last week, more than sixty of you took time to do this! Thank you!


Dept of CRISPR

The big news was that a Chinese couple gave birth to CRISPR-edited twins, a process which may have reduced the susceptibility of the baby girls to HIV. The scientist behind it, He Jiankui, appears to have broken general ethics guidelines. Three good analyses, which pick up the detail of the issue:

Sharon Begley: Amid uproar, Chinese scientist defends creating gene-edited babies.

Derek Lowe: "I do not have it in me to forgive He Jiankui."

George Daley, Dean of Harvard Medical School:

Just because the first case may have been a misstep "should in no way, I think, lead us to stick our heads in the sand and not consider the very, very positive aspects that could come forth by a more responsible pathway

but that when

scientists who go rogue ... it carries a deep, deep cost to the scientific community.


Dept of AI & tech

The five ethical design considerations of an AI society. How can we ensure an AI-society enables human self-realisation, enhances human agency, increases societal capabilities and cultivates social cohesion? Several ethicists suggest five key principles of beneficence, non-maleficence, autonomy (or the power to decide), justice and explicability. (This is a clear argument which I recommend reading, but tbh haven't had a chance to dig into it deeply enough to critique.)

💦 AI, automation, employment and demand: a clear review of James Bessen's work on how AI and automation could boost employment by increasing demand. (Before, ultimately reducing employment in a sector as demand saturates and productivity continues to increase.)

⚕️Better medicine through machine learning: a solid overview of progress in ML-assisted diagnosis, triage, discovery, anomaly detection and physician augmentation.

Deep learning has been a breakthrough technology now very widely deployed. There are still some things that aren't yet working, in domains like computer vision and reinforcement learning.

🤭 An AI-based traffic control system reported a female executive whose face was on the side of a bus for jay walking.

A lot of action in AI silicon chips this week:

Elsehwere in tech:


Short morsels to appear smart at dinner parties

Why it is easier to make decisions for someone else?

🥄 What are the best caloric restriction techniques to improve lifespan?

🤔 How real-world mood states tracked through social media behaviour can predict risky real-world behaviour. (Fascinating. And another cat-out-of-the-bag. If this relationship holds broadly, it means city or other state authorities could figure out real-world mood and sentiment of their citizens... and nudge.)

3D printed meat from plant protein, 30 minutes to print a 4-oz steak, lower emissions too. (See also, profile of Meatable's approach to cultured meat using pluripotent stem cells.)

👑 The new consumer king? Alibaba's single day hit $30.8bn in sales, more than the $24bn generated by the four-day US shopping holiday around Thanksgiving. There is a new consumer king in town. (For a more critical view of shopping holidays, please listen to my discussion with Kate Raworth.)

💚 Renewables will be cheaper than 96% of coal plants by 2030.

Fortnite added 150m users in 2018.

The mismeasure of munchkins: kids born later in the school year are more likely to be (wrongly) diagnosed as suffering from ADHD.

🔥 The WTO reckons blockchain could add $3 trillion to world trade by 2030. (Full report, which I've not read, here). Meanwhile, researchers review 43 blockchain projects in international development and find a 0% success rate.

The Taj Mahal faces an existential threat from pollution and negligence.


End note

I wanted to thank the many of you who continue to share Exponential View to your friends. It helps grow the audience for this and the podcast, this in turns gets us a more regular supply of great guests for the podcast. It also deepens the community and the expertise we can draw on.

The podcast now has listeners in 133 countries, but the audience is not as strong as it could be in the following countries Spain, Denmark, South Africa, Brazil & Israel. If you know anyone in those countries, or elsewhere, please do spend a second letting them know.

Best wishes,
Azeem

P.S. Scroll down to see what other EV readers are up to!

This week's issue of Exponential View has been supported by our partner, Hanover Communications:

Hanover helps companies decode brand reputation in the age of AI. Get your free copy of the latest report on the reputational challenges firms will face in AI revolution.

What you are up to

← PreviousNext →