December 11, 2018

Consumer AI vs Industrial AI diffusion rates - 3 Differences & Why the party is just getting started ?

All of us have steadily been exposed to a lot of media diet which advocates both the advantages and pitfalls of AI. Its the new reality. It has been silently and steadily been penetrating multiple facets of every industry . This steady state AI penetration is having tangible/intangible impact on outcomes, changing behavior, shifting business models, disrupting marketplaces. In this blog we wanted to share Fluturas practical insights on Industrial AI adoption and why that matters for the future of the economy at large which is powered to a a great extent by energy and engineering companies

At a macro level, AI Applications can broadly be divided into 2 buckets
Consumer AI Applications : Cross sell reccomenders , Sentiment Analyzers, Market Mix Modelers, Diabetic retinopathy predictors etc
Industrial AI Applications : Process chemical yield predictors, Down-hole drilling inefficiency predictors, Motor down time failure predictors while fracking etc
Question-1 : What discriminates Industrial AI adoption rates from Consumer AI adoption rates and why do those differences matter ?
Flutura found from its "from the trenches" experience that consumer AI has had a head start because of 3 reasons
  • Reason- 1 : Difference in labelled data availability for Industrial & Consumer AI models
  • Reason- 2 : Difference in perception of unlocked $ in Industrial & Consumer contexts
  • Reason- 3 : Difference in mindsets between Industrial & Consumer executives

Reason-1 : Sea of labelled data available for Industrial and Consumer AI Models
If one takes facial recognition as a problem to solve using deep learning neural nets, there is a ton of data to learn from sources like Minst.
If we take a similar problem as detecting product quality image anomalies in a diaper manufacturing plant or crack detection and progression on sub sea structures, the foundational job of creating labelled data sets need to be initiated. If the company has a "postponement of gratification" mindset the projects take off whereas if the executives want a "here a now pain balm" these projects get stalled.
What can be done about it ?
Industrial executives must be made aware that, access to labelled data will be a source of competitive advantage .
Labelled Industrial process and equipment data will be a tool for survival in the hyper competitive marketplace where access to algorithms becomes democratized and access to labelled data becomes the "moat".

Reason-2 : Difference in perception of $ value unlocked in Consumer and Industrial AI
In consumer industry when one executes for example the market mix models which change promotion $ allocation in the retail industry, executives can perceive $ value unlocked by measuring ROMI ( Return on Marketing dollar metric )
Whereas Industrial mindsets are used to perceiving value on electro-mechanical dimensions ("I can see what horizontal drilling does, I can see how adding vibration and shock sensors reduce warranty liability" which is more tangible than digital dimension ( "I cant see what I cant perceive") as a result of which they are unable to answer the question "Show me the money" with enough confidence

What can be done about it ?
Flutura has found "Engineering curves" as a good tool to manifest tangible value additions to skeptical industrial mindsets
For equipment manufacturers PF curves can be a tool for making buyers perceive value. For example Engine Anomaly detectors move detection window of warning to failure from 60 seconds to 60 mins across thousands of consumer/naval ship engines in operational deployment

For drilling contractors Depth vs Time at each rig state can be a construct for making them perceive value of AI in moving the dwell time at each drilling state unlocking millions of dollars of efficiency realization across hundreds of rigs
Every industry segment must have an engineering efficiency curve on which $ value perception of Industrial AI can be mapped. Find it and you have nailed it :)

Reason-3 : Difference in Industrial & Consumer mindsets
Because of this difference RELIABILITY becomes the operative word whether it is guarantee of prediction rates for complex fracking equipment or top drives or downhole stick shift failure.
Industrial executives carry a lot on their shoulders. A small mistake could mean in some cases life and death of humans in close contact with those risky industrial operations. Having experienced many successes and failures they need to be empathized with and gently guided into looking at operations thru new "Industrial AI" eyes.

Why do these 3 differences matter on the S curve for measuring Industrial AI diffusion rates ?
The proverbial S Curve has always been a reliable barometer to measure disruptive technology adoption rates in marketplaces be it cell phone adoption, bit coin adoption or AI adoption. If we plot the S curve for Industrial AI vs Consumer AI 2 differences in impact points will distincly emerge
Difference -1 : The take off point in Consumer AI > Industrial AI
It takes longer to implement an Industrial AI application but the potential impact of surgical AI apps can eclipse the collective impact of many consumer AI applications combined.

Closing thoughts

We at Flutura been blessed to have a ringside view of many practical industrial AI applications in the last 5 years which have massively scaled beyond "innovation POCs" in upstream drilling, process chemical manufacturing, Industrial heavy equipment manufacturers across Houston, Tokyo, Dusseldorf and a host of other industrial hubs. We noticed a difference in rhythm across both these sectors and asked ourselves 2 simple questions

Question-2 : What can be done about it ?

Lets face it ... Machine learning algorithms need to "hog" a lot of "labelled data" before the model tunes into real world behavior - be it modeling consumer behavior or machine behavior :)

" Forget the fancy terms, Show me the money" is a constant feedback we heard from Houston, Tokyo & Dusseldorf.

Industrial processes are complex with massive investments in electro-mechanical moving parts with high reliability have been in operations for many long years. Retail/Insurance/Banking process are relatively "asset light" as compared to the industrial sector.

Difference -2 : The peaking point in Industrial AI > Consumer AI

What this means is that the Industrial AI adoption party to unlock massive economic value is just getting started. As mentioned earlier there is a long horizontal line in adoption curve followed by a massive spike in adoption. We need to be empathetic to the needs of industrial executives as they manage the risk vs reward equation in a world which is accelerating and changing at a rate never before seen in human history. To conclude we at Flutura leave you with this parting thought from Jay Asher ( 13 reasons why )

“You can't stop the future
You can't rewind the past
The only way to learn the secret to press play.”