Halloween Comes To The Cloud
[Note: This is a modified version of my monthly newsletter about cloud computing. Sign up here for more.]
Sometimes the greatest truths make no damn sense. That’s not just a paradox, it’s also the definition of one.
Welcome to this month’s issue of The Speed Read, special Halloween edition.
In the spirit of “Trick or Treat” (and, really, who would ever pick trick?), I’ve assembled several paradoxes about the modern world of technology and work.
Besides being a fun treat, it’s intended to make a larger point: When you are in a new landscape, it’s hard to talk about the landscape itself. Some features are too novel to be well-comprehended. Language from the old place can’t quite describe interactions of the new. Thus, finding the truth at the heart of certain contradictions may be the best way to move forward.
Herewith, some tricky treats:
The world’s biggest computer is also the most personal.
All of the big clouds are global systems with over a million servers, cleverly networked to represent millions more computers. At any given time, millions and billions of people are diving in and out of these systems, enjoying their email, their version of the Internet, their business experience, etc. No two users are the same, and artificial intelligence at a place like Google Cloud is personalized to anticipate business needs — some configured by the user, some by utilizing agents that enable people to write things and find documents more effectively.
By comparison, servers and PCs, sold individually by the millions, delivered remarkably impersonal experiences.
In a digital world of eternal storage, vanishing analog moments rule.
A couple of years back, I calculated that 100 years ago, it cost about $30 in today’s money both to see Enrico Caruso and to buy one of his records. These days, music is basically free on YouTube or other services, and a ticket for Springsteen was recently about $1,800 on the open market. The reason why, and the reason for all the conference businesses supporting journalism and business analysts, is that as the number of digital moments has exploded, authentic human moments have become relatively scarce.
This observation brings me to the next paradox.
The jobs are going! Here come the jobs!
There is significant concern that millions of people will have their lives turned upside down by robots and AI. Maybe so; though while we have seen some robots doing relatively simple tasks, like moving things around warehouses, the impact on manual labor so far has been relatively small. And for good reason; jobs like mowing a lawn or driving a truck turn out to require a lot of contextual judgement.
Meantime, the Bureau of Labor Statistics says there are now 300,000 personal trainers in the U.S., and the category is growing at 10% over the next decade, faster than the average job growth. There are 160,000 massage therapists, growing at 26%. Marriage and family therapists, 41,500, growing 23%. You get the picture: We’re putting more money towards people who look at us, touch us, listen to us. The way machines don’t.
And that’s before we get to jobs that didn’t exist 15 years ago: drone pilot, mobile app developer, social media manager…you get the idea.
Specialize, especially on general relationships.
Elsewhere, I’ve covered the importance in the cloud computing world of software developers who are also experts in a company’s core business. That’s because products now collect information about their performance in real time, adjusting based on user demands or changing market conditions.
As AI becomes more important, domain-specific data plays an increasingly critical role in how products are built and optimized. The most successful developers must have not just the domain expertise of a specialist, but an understanding of what data around that domain matters most, how it is collected, and how to keep it free of bias. No data stands alone, and how things relate matters too.
Information is easy. Questions are hard.
A related point. AI can move through unimaginable amounts of data, finding hitherto unknown patterns and insights. That doesn’t mean the patterns are valuable: If you take all of your company’s data, petabytes of it, and demand, “Make me more money!” you’ll likely to end up with garbage. Ask specific questions, prioritized based on a company’s core competitive advantages and the best-quality data, and you’ll likely get the most useful results.
The only certainty is approximation.
I’m lifting this one for an observation from Jeff Dean, the head of Google’s research and AI efforts. He notes that advanced AI, like deep learning, doesn’t indicate decisions based on certainty, but on likelihood. Moreover, given the many layers and concatenations of a deep learning system, sometimes it’s hard to figure out exactly how the system came to its conclusion. It’s a machine that isn’t good at explaining itself, unlike, say, a car engine, which can be observed and understood with a lot of certainty.
A world increasingly dependent on statistical approximation may present a challenge to many of our existing rules, which are premised on certainty.
In an uncertain new world full of cutting-edge technology, the best advice is from a 74-year-old religious poem.
T.S. Eliot nailed our situation in “Choruses From The Rock”:
“They constantly try to escape
From the darkness outside and within
By dreaming of systems so perfect that no one will need to be good.”
No matter how much we understand and perfect the world, we’ll still have the hard work of trying to be good.