Western Railroad Discussion > Yikes - another locomotive and artificial intelligence thread

Barron's has an article this morning about the rapid growth in processing power enjoyed by Facebook and by many cloud / corporate data centers, and how that is leading toward advances in "inference". Inference allows the machine to review thousands, millions of past events and infer what might be happening now, and what is the best course. Not unlike a rookie engineer sitting with an old head who's worked the line for 40 years. He's seen just about everything, and often senses what is happening.

The article is focused on what's happening in the technology side, with the companies who build the processors or storage capability. But, it does suggest what could be happening with the billions of hours of downloaded data files from locomotives. Matched with geographic data, and I believe most locomotives have and record that, characteristics like throttle position, reaction time to changes, load characteristics and DPU reaction, etc can be replicated.

Unlike algorithms, which are essentially programs written by somebody or something, inference looks at what has actually happened in the past, and evaluates possible outcomes in the present. Managing slack, working the fence, mixed vs unit train consist. At this point in the route, how do the best practices direct me to act? Building a train for maximum handling security, taking on the fly inputs like radar determining a clear track, and signals that crossing barriers are working as intended, etc.

It's interesting stuff to read and ponder. Probably still a ways off for railroads and airlines, but likely within the lifetime of most people reading this post.

[Barron's is often behind a paywall, but some content is released on Monday. Or search ai technology revolution tiernan ray]

https://www.barrons.com/articles/ai-technology-revolution-is-just-getting-started-1527292801

I haven't seen anything contradicting what I know about AI research, even now, that AI is essentially an imitation of how our brain is "supposed to" function. Keep in mind that even experts don't understand how our brains really function.

The saying "imitation is never good as the original" really rings true here.

While I'm not dismissing the research and the progress they're making, there are several factors that make hyping of the technology more imperative than ever. The recent setbacks by self-driving vehicle accidents (that includes Tesla, which is not "truly" self-driving) need to countered with positive stories like this. However, there's a far bigger problem that might crush the financial health of this industry.

China is also involved in research on Artificial Intelligence, and they are spending large amount of resources now. We are supposed to be the pioneer and "the monopoly" to export this technology in order to get the financial returns from the massive amount of investments already spent. A formidable competitor is going to ruin that scheme.