IoT AI :: AI IoT
In the land of self-fulfilling anomalies you learn to see through those high on the possibility, to focus on those exhausted from the probability. The probability they are going to build something valuable, of worth, something that people and companies want, that they HAVE to have. Something that stands out no matter what, literally at [half] the speed of light as it transcends thru those fiber-optic lines as fast as the web masses can authenticate and download. That type of data don’t lie. Doesn’t it?
It doesn’t. It seems to work pretty well in this current state of software, an unmistaken feedback loop that many have experienced if not figured out. Nothing wrong with an unbroken system, except it’s not the one that will replace it. I have to admit, it does feel like we’re on the cusp of something new. So many devices, so much data; yet so distracting, it’s even overwhelming at times. Not that the connections are necessarily getting out of hand [I believe all should have the power of a smart-phone if they choose], but the WAYS we as a species/civilization interact with the internet’s resources is about to change drastically. The shift is set to be a more passive, less consuming interaction and experience in our ‘natural’ environment. Per capita, connected devices are on the verge of exploding, and yet I believe [and hope] that IoT represents a distinct opportunity for many to decrease their overall #screentime. I cannot deny that AI’s taught to learn how to decipher and interpret all of that IoT data makes more sense than we probably know. Which is the point I guess, more than we can know/process in the infinitesimal time that such data remains optimally relevant. How are we going to realistically make sense of that much data fast enough to justify the cost otherwise? So let’s say for a sec that fully utilizing various IoT capabilities requires proportional investment[s] in AI, a proportion yet to be determined, but nonetheless. Following me? Watch me now, this gets interesting with the quickness.
If IoT necessitates investments in AI, and an AI is programmed/equipped to then crack that pattern pretty quickly (pattern being: that ‘self-investment’ to some quantifiable extent fulfills more IoT capacity and capabilities), then well now … THAT is a SMART system! At a specific, adjustable ratio the program more-or-less runs: ‘boost me, boost objective’. Not that we have pushed deep learning quite that far yet, but what if the system can be aware if/when its performance is the limiting factor, rather than optimizing explicitly external variables? You could program the system to get ‘smarter’ by improving itself as a function of decided IoT outputs. Hence, IoT AI :: AI IoT, or at least that first part. AI IoT would refer to our world expanding its internet connections under such scenarios where their creation/installation reflects back into the AI recommending them, and thus the loop is closed.
You could program the system to get ‘smarter’ by improving itself as a function of decided IoT outputs.
Now let me see, who is investing immense amounts of money into Artificial Intelligence? Let me think. Starts with Goo-, ends in -gle. My hypothesis is that Google is investing in AI to dominate IoT, or more specifically, to dominate certain components or categories of what we currently classify as IoT. At any rate, this computes when you look at portions of the IoT progression as a natural extension of the global ad network that Google practically owns in many markets. What do you think? It’s easy to see how the two trends/phenomenon are related. Sure, in the beginning it will be mostly sensors sending back data to software that then reports/interprets it for human decision-making, but the real opportunity will be in equilibrium systems that optimize to maintain/minimize/maximize various criteria (i.e. equations). If I’m right, Amazon beware. Google will most certainly be coming for not only the advertising, but perhaps a piece of the transactions its AI recommendation-engine(s) make (an auto-fulfillment feature I can already visualize).
So here’s a theoretical use-case:
- AI monitors an IoT network of some particular type.
- Factors plotted/mapped/forecasted suggest a certain [high] probability that X number of replacement hardware components will be needed.
- Software locates/purchases replacements, confirms [auto-]fulfillment.
- Installation crews deployed.
- Downtime of partial network utilization minimized.
The key differentiator between that scenario and how most are characterizing the onset of IoT is in knowing to some confidence interval that those components will be going offline before it happens. The standard upgrade is thought to be in the fact connected sensors will effectively ‘report’ their malfunction without manually checking on-site, but without stretching too far down the inference extension, a ‘smart’ system could be taught to identify based on prior datasets certain ‘fail’ scenarios.
I believe that the predictive capabilities of the IoT/AI symbiosis will be a core driver of the onset/investment in both.
IoT provides data, AI interprets it. And then provides recommendations, and in some cases is programmed/approved to take action(s) itself, and most certainly the AI learns/improves as a function of all the sensors/networks it has access to. Pattern recognition and modeling in our Earth system is going to change forever. ‘Forever’ to the extent the collective internet system doesn’t forget, but when some of these technologies hit the open public markets, the incentives to reverse engineer alone will be sufficient to warrant the word ‘forever’ (on a human time scale at any rate).
Regardless of whether this little thought exercise bears any actuation, there’s a wide world of world-wide possibilities coming up. These growingly-common terms: orchestration, automation, optimization … predictive capabilities, are all set to grow exponentially in such a landscape/scenario. Let’s just hope those AI systems directed to analyze photosynthesis don’t get any ideas. 🙂