My brother-in-law, Mark, has a plaque on his wall that says, “In order for artificial intelligence (AI) to evolve from zero to Einstein, it first has to go through Larry, Moe and Curly,” for those of you that recall the iconic Three Stooges. It is difficult to assess, in real time, where we are on the evolutionary path, but it appears to be approximately the point at which salamanders grew hair.
The utterer of that ageless quip, which cannot be ageless when AI remains only a fantasy, was, immodestly, me. Mark and I were chatting on vacation; he a safety software expert for light rail systems in Eastern North America, me a reporter of passingly interesting stuff.
My grandson, Adan, a consumer of passingly interesting stuff, seized upon my recent purchase of a software suite for photo enhancement called Topaz AI, the AI being for artificial intelligence. To Adan, the mocking of AI and the purchasing of AI marks an incongruity that demands explanation.
It’s not that hard. Calling something AI does not make it so, and neither does grabbing a fad name to market a product make the product inferior.
In the case of Topaz, the product is not inferior. I shoot a lot of pictures. Those of you that got our Christmas card saw an image of the Credit River from my back yard shot with a 14 mm lens at dawn.
Since I shoot quite a bit in low light, I sometimes use a high ISO setting so I can keep my shutter speed fast enough to stop blur and my f-stop high enough for the lens to work. The problem is, shooting at high ISO causes digital “noise” — little flecky things that soften the image and muddy the colour.
Topaz AI Denoise uses machine learning, aka artificial intelligence, to figure out how to replace the flecky things with colour information. The other products in the suite, Sharpen, RAW to JPEG, etc., also use machine learning.
Except it’s not a machine. I get it. It runs on a machine. And it’s not amazing that digital products need to learn. In the old days of MS-DOS, you could let your programs run along on their tracks until there was a crash. That’s what they called it — a crash. And then you had to start over, reboot, take out a conflicting command string and run it again. There was nothing artificial or intelligent about it. Back then, it was called “logic,” but, as with all things digital, it was not logic or real. It was what it was in its digital world of high science and Pac-Man.
Machine learning started off as software plagiarizers not wishing to create their own, brand-new game or science, so they stole somebody else’s, called it “open code” and built on the existing foundation with no thought of sharing credit or revenues. Not, that is, until somebody stole their code, built on it and left them in the dust. There is a whole global community out there of frustrated, erstwhile software moguls watching better thieves eat their lunch. You live by plagiarism; you die by plagiarism.
It’s a bloodbath out there. I see in last weekend’s news that Canadian luxury-coat manufacturer Canada Goose is bending to the will of the woke culture and apologizing for the fact that feathers and fur made them rich. Now, the only “cool” Canada Goose apparel is either used or fake. Using real fur or real down is not good. This is especially true for woke machine learners that need heavy arctic gear for a subway ride to Manhattan. For people that need to stay warm in the Arctic, you need real fur and real down — 800 fill power, minimum, and when I was winter camping in Canada my source was typically Seattle-based Feathered Friends. Even then Canada Goose was too urban for what I needed. You have to match the technology to the application.
Another old adage we all know from our infancy in business is that there is nobody easier to sell than a salesman. AI is all the rage today, just as LinkedIn was to be the end-all in “networking.” Or MySpace was to be the social media pinnacle. Or whatever.
But we learned that, while the best way to debug a software was to let it crash, we needed to step beyond the machine before we ran a test on public rail systems. Instead of fake intelligence, we needed the real thing.
A few years back I reported on a fraud I covered back in the ՚90s. A guy was claiming to infuse “DNA” into a “chip” to help people find bombs in Afghanistan. The gimcrack was called a Quadro-Tracker. You can find it with Google, or you can search with real English and real intelligence on our website for the story. DNA, like artificial intelligence, was a sham built on a reality. A hoax. A sales tool for the weak and infirm.
People ask me from time to time what this or that has to do with the wood industry, after all. Especially, how will this information help me sell?
For most of my friends reading this and working to help the wood industry, it is, has been and will be obvious. For the other few, riding on artificial intelligence and DNA …. Well, they would be more entertaining if they weren’t a runaway train in a populated industry.
For some real, production-focused networking, I hope all of you will see Steve tomorrow and the rest of this week in Salt Lake City for this year’s WMIA/WMMA Wood Industry Networking and Snowsports Event. Real skis, real snow, real people and real friends.