Machine Learning, AI and Beyond— What are the Limits of the Hype?

As data science tools have enabled machines to do complex analysis, predictions, and transformations off of not just extremely large text data sets, but visual and audio media, machine learning has become an emerging hot topic. What was once sci-fi, has become a reality through advanced computational capacities of machines, and massive sets of data available online.

With image processing, aural processing, and natural language processing techniques being propelled forward by machine learning and deep learning (neural network) algorithms, there’s a ton of hype around machine learning, deep learning, and “Big Data” or “Data Science” right now, riling up a lot of speculation about what machine learning will be able to do in the nearer and further future. Many in the realm of data science may disagree — often on the basis that to naysay now can pull some of the massive, multimillion to multibillion dollar funding AI projects are recently graced with — but, applications of deep learning pushed on anything and everything theoretically feasible as a business application of AI algorithms, can be massively computationally inefficient, and simply unreliable and inferior to more straightforward machine learning and human led analysis and research methods. As with lots of hot tech, early on, there’s tons of inflated expectations — human error in understanding what technology is capable of in the short term without longer development cycles, or assertions in bad faith propagated to excite investors or otherwise impress people.

Eager investors being taken for a ride.

This post was largely inspired by a conversation at a data science panel I attended, where I posed a question about this amusing, but quite insightful visual I encountered about a year ago independently:

A similar pattern to the dot com boom and bust has been seen recently in the blockchain space, on the graph being depicted as having just past peak hype and on it’s way to the Trough Of Disillusionment, though recovery of cryptocurrency markets may indicate this cycle was particularly short, perhaps showing that blockchain has passed this trough already as it is on a trajectory now towards mainstream adoption. Of course, only after numerous failed blockchain projects and cryptocurrency scams lay in it’s wake — again, not dissimilar to the bust of the dot com era.

One of the leading thinkers, and vocal critics, of the world of AI is Gary Marcus, who has a background both in artificial AND natural intelligence research, credentials a wannabe neuroscience researcher turned data scientist like me thoroughly respects. If we are to create true general AI, how can we achieve that without thoroughly understanding the neuro-electrical machinations of our own natural intelligence? Gary agrees we must relate AI to natural process if we are to get there. No matter how impressive machine learning and deep learning models are at crunching more and more computationally complex data, now able to recognize faces to create all sorts of amusing filters people use on social media, segment photos and genres into useful categories for consumers and businesses, and, on the scarier side of things, creating deep fake videos that will be more and more difficult to discern from the real deal, deep learning does not actually know the ins and outs of complex social interactions between businesses and clients, it cannot create moving works of art and music from scratch, it cannot match the creativity of arts and language employed by even young, barely verbal children.

As we see in our hype cycle chart, regardless of how powerful a force AI will be in coming years, in the short run, the hype will fall short, painfully expensive mistakes will be made, companies will evaporate like flashes in a pan, and there will be a time of uncertainty in the sector when advances in AI research will stall and sputter, perhaps at it’s most nightmarish, we’ll have just enough development to propagate deep faked media further, while dreams of utilizing AI to solve great issues of socioeconomic inequality, and ecological disaster, will be muddled with panic as some parallel global crisis emerges as AI disillusionment mounts.

There is nothing unusual about this now all the more predictable pattern. You can be assured, that the world of AI, as with other emerging tech, will quietly and quickly make massive strides maybe a year or two after it’s impending, most uncertain or darkest moment. I find that this is poetically analogous to a sort of emotional-spiritual crisis that many people face at some point in their lives, a Dark Night of the Soul, which, no matter how painful and insurmountable it may seem to the afflicted at it’s worst moment, is temporal, and often signals profound, meaningful transformation impending, perhaps a transformation we couldn’t even imagine prior to our hopes and dreams smothered by this darkest of moments.

Image result for darkest night sun

Operating at the intersection of Technology, Ecology, and Arts

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store