Inside | Real news, curated by real humans
Inside AI

Inside AI (Aug 18th, 2019)

Hi Everyone and welcome to the weekend edition of InsideAI.  Please pass it along so your friends can sign up.  I'm Rob May, CEO at Talla, and an active AI angel investor.  I also run the AI at Work podcast.  If you are new to the newsletter, the weekend edition is about summarizing the top articles from the week, and providing some commentary on things I'm noticing in the AI ecosystem as an entrepreneur and investor in the space.

Here are the most popular articles of the week:

Tech entrepreneur Bruno Aziza recommends these three books for brushing up on your AI knowledge. They are "The Big Nine: How the Tech Titans and Their Thinking Machines Could Warp Humanity," by Future Today Institute founder Amy Webb; "Only Humans Need Apply: Winners and Losers in the Age of Smart Machines," by author Tom Davenport; and Stephen Hawking's perennial "Brief Answers to the Big Questions." - FORBES

The Allen Institute for Artificial Intelligence and the University of Washington created an AI that can generate fake news based on only a headline. Grover was trained on 120 gigabytes of news articles to actually spot fake news written by AI, but its developers also taught it to generate false articles as well. In scorings, people rated it as more trustworthy than human-written false news. In his own experiment, Fast Company's Mark Wilson put in the headline: "Why Donald Trump Eats 100 Cheeseburgers a Day.” The results, which are here, are pretty convincing. Grover is also online for anyone to try. - FAST COMPANY

Several current and former employees of SoftBank-backed claim that the startup exaggerates its tech expertise and AI capabilities. The news was first reported by The Wall Street Journal, which notes that the accusations "reflect a growing challenge in the tech world of assessing a company’s proficiency in artificial intelligence." Last year, raised Series A seed funding of $29.5 million, led by Lakestar and Jungle Ventures. The startup uses a human-assisted AI to help companies build custom-made software. - WSJ

British artist Anna Ridler collected and photographed 10,000 tulips as part of a dataset and artwork featured at the “AI: More than Human” exhibition in London. Ridler visited the Netherlands to gather the tulips, label each picture (noting its stripes and colors), and then used the dataset to create a 1,614 square-foot installation, comprised of the images, titled Myriad (Tulips). In addition to the artwork, Ridler, a self-taught coder, fed the photos and labels into a machine learning model to create a three-screen video installation. The model is linked to a real-time feed that reacts to cryptocurrency prices. “As the price of Bitcoin goes up, the tulip becomes more stripy,” Ridler says. “As the price of Bitcoin goes down, it starts to become more plain." - BLOOMBERG

A few weeks ago I wrote about why automation necessarily won't kill jobs.  I've already seen some evidence this is playing out, while talking to some friends who recently tried the Stitchfix competitor Trunk Club.  If you aren't familiar with Stitchfix, it's a personal stylist clothing service, run by algorithms and overseen by people.  You get a monthly box of clothes that the algorithms suggest.  Stitchfix used machine learning to dramatically lower the cost of personal stylists.  But then what happened - we got more personal style services (and I assume more personal stylist jobs).  Why?  There are three reasons.

1.  As I wrote in the original post above, some people rebel against the algorithms and consider it a luxury to have a human do the thing the algorithm did.  It's a sign of status.  Having a personal stylist used to be only for the wealthy, Stitchfix brought that to everybody, so, having a human do your styling is a status symbol.

2.  Some people don't trust the algorithms, and humans sometimes need to be the interfaces to other humans.  Personal style algorithms can now make many mediocre personal stylists into really great personal stylists, and for those people who would rather not believe an algorithm suggested their clothing but can't afford a high end stylist, the combo model using humans as the final interface works.

3.  The algorithms still need humans to improve, and catch new trends and ideas.  Machines are limited to learning from large data sets (at the moment) and can't do things like counterfactual thinking or other types of analysis that might shed light on fashion trends.

AI is going to lower the cost of doing many things, which means we will get more people doing those things, and need more jobs to service those people.  As I've said before, maybe at some point in a few decades machines beat us handily at everything, but, for the next decade, I am very bullish on AI creating more jobs.

Thanks for reading.


Subscribe to Inside AI