award show music…it’s the CLOUDYS!!!
Or even crazier, the second annual Cloudys. The last email of each year we throw an event called “The Cloudys”, an award show dedicated to excellence in AI/ML, cloud computing, a little blockchain and anything else that I feel should be awarded. The only tradition of the Cloudys is there are no traditions, except for one, and that’s at the very end.
Since the event isn’t live, just imagine celebrities reading the categories and handing out little silver cloud shaped awards to the winners, people crying, thanking their mom, dad, God, all that jazz. On that note, let’s just get started.
Technology that was amazing when first announced, now is a commodity:
Remember when Amazon first released a video about their cashier-less store; it was amazing and I was lucky enough to go to the Amazon Store this year. Now apparently it is a dime a dozen. Sam’s club has plans to open one, as does French based grocer Casino, and there are plenty of startups in this area. While Amazon’s store seems more sophisticated in its use of AI/ML, others will probably copy the technology. It really is amazing how quickly technology gets copied.
Biggest AI trend where people still don’t know the timing: Autonomous cars
Remember when autonomous cars were right around the corner? I believe 2018 was a mixed year. The positives were Google continues to collect millions and millions of more miles, and self driving car valuations are getting bigger, but that is about it. The negatives are also big, with Uber’s unfortunate fatality, which was clearly the fault of their machine learning, to drivers complaining about how much they hate the cars on the road, and finally Tesla removing the “Full Self Driving Capability.” Even more concerning, is some chatter on Twitter that some people believe solving autonomous driving is closer to solving artificial general intelligence, which we are nowhere near.
Now, Google is actually planning to launch their autonomous service; the catch is it’s only in a 100 square mile area in Phoenix and will only be available to about 100 riders. Google has spent over $1bn on Waymo, thus if 100 riders are in the program, it would mean their customer cost per acquisition was $10mn and if we assign each member an LTV of $10,000, their CAC to LTV would be 0.001, three thousand times smaller than the ideal ratio.
That was a joke.
Which brings us to our next cloudy…
Most ridiculous survey: MIT Media Lab
People love to apply some version of the “Trolley Problem” (an ethical question involving whose life you would save if a trolley was about to crash into different groups of people) to autonomous cars, often invoking ridiculous situations.
The MIT Media Lab has taken this initiative one step further, by polling over two million people about absurd autonomous car scenarios. This is a great example, where respondents are asked essentially to choose if an autonomous car should crash and kill either “2 girls, 1 man, 1 woman and a boy”, or “1 women, 2 elderly men and 2 elderly women.”
Using the data, researchers then rank who respondents would spare the most. Not
surprising, babies would be spared most of the time, but if an autonomous car
was going to kill either a dog or a criminal, more people would choose the criminal to die. It’s
not fair to the criminals, because we don’t know if they committed tax fraud vs
arson, or have a DUI or killed someone,so the second study should really
account for this.
Preferences in favor of sparing characters
…and to all of the cat lovers out there, the data speaks for itself
Trend that should be made into a movie: Bitcoin mining
The movie starts with a plane flying over the tundra in Canada. All you see is the barren wilderness, covered with snow. As the plane speeds forward, in the bottom left of the screen, text comes up:
Medicine Hat, Canada.
Then, off the in the distance, a small black cluster can be seen. As you fly closer, you see row after row of black shipping containers which seem to go on forever. The plane circles the cluster, and then lands in the snow. Four men get out. They are greeted by a man who brings them to the largest of the clusters, and opens the door. Here you see rack after rack of computer equipment. The title of the movie then flashes:
While slightly dramatized, this actually, sort of happened. Yes, there is a real place in Canada called Medicine Hat, yes, they have their own newsite, yes, Bitfury built a $100mn server farm to mine bitcoin there, and yes, people leased Boeing 747s to fly Ethereum mining equipment to mining facilities.
The real life stories of Bitcoin mining are insane, from China banning mining operations, causing miners and exchanges to move out of China in the night to the Venezuelan police confiscating mining tools. You could certainly make a movie about some crazy international plot to shutdown Bitcoin that involves the destruction of these facilities and then an unlikely hero saves the day, or terrorists could take over the facilities and cause financial destruction; whatever it is, this movie would be amazing.
Or, at least a book would make me happy.
Most disturbing trend of 2018: Companies claiming to be “AI” companies, yet have no one who has a background in AI working for them
First off, I am not expecting every company that gets into AI to have a former PhD of machine learning as the CEO; nor would I want that. However, what I have started to notice is companies claiming to be an “AI” company, yet have no one who knows anything about AI working for them. Trust me, when I see these companies, I scroll through every single person on LinkedIn that works for the company and try to pull something, I mean, anything related to machine learning. Be careful out there.
Strangest Google Trend Finding
When you search “Artificial Intelligence” in Google Trends, Blockchain is the top ranking “related term”, followed by “sex dolls” and “dolls.” The world is a strange place.
I was traveling back from a client meeting, waiting for a bus to take us to the airport. As we were leaving the parking lot, I saw this particular company had installed several electric vehicle charging stations and then I thought to myself, “OH S***. This is how the electric revolution comes to play.” Instead of their being a few gas stations where everyone goes to fill, electric charging stations allow people to refuel their car anywhere. In 20 years time we may see a world where there are electric chargers everywhere and people can just seamlessly charge their cars and it starts in the office park.
Best article that describes how AI is still in its infancy
This is the best AI/ML article i read all year.
With the best quote of the year:
As one competitor put it: “Everyone starts with machine learning, and eventually, everyone realizes it doesn’t really work.”
THE FINAL CLOUDY – The short sighted cloudy
Each year we end the Cloudys with a story that represents, terrible short term thinking. Last year was a story about how in 1999 McKinsey estimated mobile phone sales would never top 1mn per year; their reasoning being they were big, had short battery power and were expensive to operate.
This year we feature a story from “Master of the Senate”, Robert Caro’s third book in the series of way too many books on LBJ. Caro begins the book by talking about the history of the Senate, and has this great passage:
“Senators of the United States grappled—as, once, the senators of ancient Rome had grappled—with the concerns of expanding empire: should the borders of the young republic be extended west of the Mississippi, and if so how far west—to the Great Plains, or even further, to the mighty mountain chain of the West and the shore of the great ocean beyond? (Many senators considered this last suggestion ridiculous. When, in 1824, there was a proposal for the erection of a fort on the Pacific shore of the Oregon Territory, Mahlon Dickerson of New Jersey said there was no realistic possibility that Oregon, separated from the United States by virtually impassable deserts and mountains, could ever become a state; even if its congressmen managed to cover twenty miles a day, he pointed out, they would need 350 days to get to Washington and back.”
A year after Dickerson made this statement, the first railroad was finalized in the US and while only 9 miles long, was a hint of things to come. Eight years after that, the first railroad from Baltimore to DC was finished; another 15 years later there was more than 9,000 miles of track laid; and 19 years later, in the middle of nowhere in Utah, the Golden Spike, or last spike of the transcontinental railroad was completed, making it possible for someone to travel from New York to San Francisco in a little over a week.
Dickerson passed away in 1853 and didn’t see the completion of the railroad, nor the benefits it brought. But, don’t worry, what happens to be about 1000 feet from where Dickerson is buried?
See you in 2019