A Blog by Jonathan Low

 

May 20, 2019

How 3D Printers Are Making Better Body Parts

With greater precision. JL

The Economist reports:

Because 3D printing lays down an object layer by layer, complex shapes with intricate internal structures can be built to print a special porous surface onto implants. That surface encourages bone to grow into the implant, which secures it more firmly in place. When combined with the precision of robotic surgical processes, this makes replacements more successful. The medical and dental use of 3D printing was worth more than $1bn in 2018, 11.5% of the entire market in 3D-printed goods and services.

No, AI Is Not Going To Solve Facebook's Problems. And Maybe Nothing Is

Facebook's mantra has been that technology will solve the problem's it created. Most knowledgeable - and technologically unsophisticated - people outside the company are skeptical. Turns out even those inside the project acknowledge its limitations.

 The larger question is whether there is a cure outside of changing the company's business model. JL


Cade Metz and Mike Isaac report in the New York Times:

The social network has been under scrutiny for the false, misleading and inappropriate content people publish on its site. Mr. Zuckerberg testified Facebook was developing machine-based systems to “identify bad activity” and “over a five- to 10-year period, we will have A.I. tools” that can detect and remove hate speech. He has repeated these claims with the media, with Wall Street and at Facebook events. The question is whether that is true. Identifying rogue images is one of the easier tasks for A.I. It is harder to build systems. Behavior changes. Attackers create new techniques. The person at Facebook leading the effort acknowledged that A.I. alone could not cure Facebook’s ills.

How Automation Is Changing the Workplace

Companies seem to be trying to use the new technologies as a way of enhancing productivity, rather than simply saving on personnel costs. Over time it may reduce the number of workers required both blue and white collar (see Ford's 7,000 employee layoff announcement today).

But the early experience suggests that those combining technology with improved human training will reap the best results. JL


Ezekiel Minaya and Tatyana Shumsky report in the Wall Street Journal:

Software reduces a lot of the labor hours needed. The other benefit is that the remaining work is more rewarding. They’re doing more analytic work.You do need to provide training. We begin to plan what areas of emerging work we want them to move to, and if their skills are not aligned there, we’ll put together training. Then there has to be an effort to get all of the managers and operators and department heads to embrace the idea. If you do that, employees are not as worried about digital transformation.

Why Media Buyers Are Mixed About Selling Emotion-Based Ads

It's one more data point in an increasingly crowded and difficult-to-verity market full of new and unproven metrics.

What may be most interesting is that this appears to be aimed as much at reassuring marketers their ads are actually targeted at potentially monetizable audiences than at dramatically improving sell-through. JL


Sarah Jerde reports in Ad Week:

(The publication) predicts how its stories make a reader feel using machine learning and asking readers how they felt after reading it. The publisher then sells ad space around those stories based on indicated emotions.While some consider it to be one more data point to target a specific audience, skeptics call it another type of contextual advertising that’s too abstract to rely on. Buying a particular emotion is usually sold as part of a broader package. This gives buyers another level of security that their advertisements will appear alongside appropriate content.

Dont Blame the IPO Bankers: Ride Hailing's Structural Economic Inefficiency

What became evident in the course of watching Uber and Lyft go public, then seeing their newly minted stock prices decline, was that markets, whatever their issues, have become pretty good at pricing risk, especially in technology-related companies.

The problem is that the companies will lose customers if they ever contemplate charging profitable prices. And driverless cars will only make their problem worse by reducing barriers to entry without necessarily reducing costs. JL


Lyall Taylor reports in the LT3000 Blog :

Few people are able to afford to pay someone $10-15 an hour to drive them around when they can do it themselves at no additional personal time cost. Private chauffeurs are rare for the same reason private cooks and private butlers are. It's an economically inefficient use of human labor. Because ride-hailing is economically inefficient, the only way to generate mass-market demand is by underpaying drivers and underpricing its services. The coming of self driving cars is an existential threat to whatever residual of a viable business model does exist. Far from being the future of mobility, Uber is on the eve of being technologically disrupted.

Should Companies Use AI To Assess Job Candidates?

AI proponents argue that the traditional, totally human hiring system is idiosyncratic and inefficient.

They tend to ignore the fact that over the many millenia of human existence, highly functioning, productive people and organizations have somehow been identified and gainfully employed which has resulted in generally improved socio-economic performance.

They also tend to gloss over the concerns raised about misuse and abuse, instead offering platitudes laden with lots of 'shoulds' rather than 'musts.'

The reality appears to be that AI and algorithmic assessment can aide humans in making better judgments. But smart organizations recognize that using AI as a substitute rather than an assistant - often for reasons of cost and scale - may simply result in a new set of problems rather than an ostensibly ideal solution. JL


Tomas Chamorro-Premuzic and Reece Akhtar report in Harvard Business Review:

There is a difference between what we can know about people, and what we should know about them. Humans can accurately identify personality and intellect from thin slices of verbal and non-verbal behavior. AI algorithms leverage the same cues that humans do. The difference between humans and AI is that the latter can scale, and can be automated.AI mines voice; body movement, or facial expressions to predict personality, one of the leading predictors of job performance. Similar signals predict communication skills, persuasiveness, stress tolerance and leadership. (But) it is advisable for candidates to have ownership of their data and results, which they may voluntarily decide to share with employers - or not.

May 19, 2019

Rise of the Online-Only Relationship

What is virtual, what is real - but more importantly, does it matter anymore? JL 

Christopher Mims reports in the Wall Street Journal, illustration by John Kuczala in iStockphoto:

The percentage of Americans age 18 to 29 who report not having had sex in the past year was 23% in 2018, where in the early 1990s the figure was half that. Many of them spend months or even years dating without ever meeting face to face. The technologies that make it easy to connect with others all over the world have yet to give them the ability to teleport their bodies. It’s possible to find someone who happens to share one’s particular combination of tastes. It’s one of the ways teens and twentysomethings are adapting to a combination of two demographic trends—earlier puberty and later marriage—using  technology.