Things to watch out for when implementing AI in your organization

Almost 62% of respondents in a 2017 business survey pointed to Artificial Intelligence (AI) and Machine Learning (ML) as the two most important factors that will set the tempo for business for the next decade and beyond.

Not to be outdone, Big Data and Business Analytics (BA) came in right behind at 58%.  Don’t fret about the totals exceeding 100% since participants were asked to rank dozens of factors in order of importance, likelihood, or need, and choosing one thing does not preclude another.

AI Expectations

AI is coming, and few are even thinking about applying the brakes.  There are so many massive advantages to implementing AI, from getting a leg-up on your competitors, to asking “just the right question” from your data to revolutionize the world.

The key thing to focus on is that though this technology can hardly be considered “new” anymore, its relevance and importance are growing on a daily basis.  There is not going to be a business anywhere in the world that doesn’t incorporate AI at some point, and that brings (and requires) a whole cascade of new technologies and techniques for collecting, collating, manipulating, and analyzing data.

 

What could possibly go wrong?

Of the respondents, 95% are convinced that AI will simplify their job, eliminating a lot of the tedium, allowing them to get more done, faster, and freeing up time for them to be creative.  On the flip side of that coin is the faction who think that just asking the foregoing question could bring on disaster.  We do not have time for these superstitions minds.

Ignorance may be bliss, but it doesn’t change the fact that this is the question we must ask if we’re going to prevent the most fundamental problems from crippling our efforts.  This is not a time for blind faith.

The enthusiasm by the upper echelons is palpable for AI, but so is fear and apprehension on the part of workers who will be affected.  Will it take my job?  Will I be relegated to the scrapheap of history?

The Evolution of Work

Every time a new technology is introduced, people are displaced, but historically speaking, new jobs (often better paying) replace older jobs.  Look at the mining industry, moving from hand labor to automation.  There are more jobs than ever in that industry, but now people are technicians and specialists rather than hewers of rock.

Speaking of which, when it comes to data, mining has both new aspects and interesting parallels.  Data mining used to be done entirely by hand, going through tangible paper files, extracting tidbits and making a new paper document.

Google is in the process of turning the 130,000,000 existing unique printed books (non-electronic) into digitally searchable electronic documents.  They are using Optical Character Recognition (OCR) and vacuum-powered page-turners to automate the process.  Soon, everything we’ve recorded, in all languages, will be available in a digital format.

The Questions You Need To Ask

AI will eventually permeate every aspect of our daily lives.  The questions to ask are: is it beneficial yet; is it practical or a gimmick; and how quickly should you get onboard?

Gimmicks

In business, gimmicks aren’t necessarily a bad thing, especially if they grab customers’ attention, endearing them to your service or product.  Done right, you can have a customer for life; implemented poorly you could lose them forever, possibly decimating your company.  Being “cool” is one choice, but really, genuinely improving someone’s life is the preferred route.

Business Case

Can you make a business case for AI?  Artificial Intelligence is not a stand-alone product.  It is best suited to work in conjunction with your existing system and people.  It is a complementary technology, not a whole-cloth replacement.  The most common failure of AI implementation is when a business leaps onboard without any idea of where they are going with it.

Where is it Useful?

ChatBots are particularly worthy because psychological studies have shown that customers start to get angry when they are put on hold for more than 17 seconds!  This is particularly true with 100% automated systems where there is no human contact.  Humans dread “Your call is important to us and has been placed in priority sequence…”

ChatBots solve the problem.  Some are now so sophisticated that they are indistinguishable from a human being.  As long as you have the data bandwidth, a single ChatBot can converse with hundreds of customers simultaneously (great for power utilities when there is an electrical outage, for example).

But it’s also equally efficacious in your Accounting Department, where it can handle much of the daily transactional business, or alongside any business function where it can take over the tedious, infinitely-repeated steps that cause employee burnout.  AIs don’t get bored or tired, and thus don’t make “bad judgments” or mistakes as a result.

Reliable Data Sources

GIGO, or Garbage InGarbage Out, is a very old, but profound expression in computing.  If you don’t have a massive company database and reliable data from which to extract insights, relationships, and parallels (the bread-and-butter of Artificial Intelligence), a custom-designed, in-house system is not going to be very useful to you.

 

Don’t panic, however, because that doesn’t exclude you from playing the game.  You could certainly enhance your database, but there may be a better alternative.  You might be better off with AIaaS (Artificial Intelligence as a Service) where they could collate all your data for you in a high-security environment, with the added benefit of it being off-site and possessing backup copies.

Your Hardware Stack

Also, it doesn’t necessitate having to change your hardware environment to accommodate an AI System.  You should also ask yourself if your current software stack is complementary with AI.

 

Indeed, should you buy a new $100,000 rack to sustain an AI that must be serviced, maintained, and constantly updated by an on-site staff?  The cost of that single rack is the same as the salary for one Data Scientist for a year, of which you might need half a dozen.  Many companies are looking to AIaaS as the solution.

They provide their Data Scientists, managing AI for dozens of companies.  Where it is nearly impossible to find a graduate in Data Science nowadays, this “sharing” is providing a bridge until there is a fresh crop from which to hire.

Useful, but Not For Us

Does a flower shop need a ChatBot capable of handling hundreds of customers simultaneously?  There are many places where it is simply not warranted.  Referring back to the Business Case scenario mentioned above, it may be possible—it may even be cool—but there are some situations where it’s simply not justified.  Be ready to use your common sense when enthusiasm starts to take over.

Not a Panacea

Pretty soon we’re all going to have reasonably sophisticated AIs built into our smartphones.  That will be more than enough for some people running small businesses.  There is, however, another seldom considered possibility.  AIs analyze data, and they can be wrong!

The result might be due to a bit of bad data, or it could be completely good data with a poorly phrased query—that doesn’t matter.  What does matter is oversight and not subscribing to blind faith?

 

AI is not suitable for every task—not in its current state of development—and you have to know where to draw the line.  If something seems a little off, or plain wrong, it is up to you to demand more information.

Transparent Logic

You need to be able to see how the AI arrived at its assessment.  AI is not some magic trick where you want to maintain the mystery.  When called upon, the AIs should be able to show you step-by-step how it arrived at a particular conclusion.  Accept no “black box” AIs, and keep your Critical Thinking honed and ready for use.

Employee Fear

The modern public is anticipating wide-scale AI because of the perceived advantages—talking houses, self-driving cars, et cetera—but when it comes to their jobs, they worry.  It is essential to communicate with your staff members and show them how AI will benefit them, making their job more manageable, and more rewarding.

 

It is an ongoing process, and they should be involved right from the start.  Assuage their secret fear that the company will implement an AI system and one day say “Okay, you’re all fired.  The machine can do it all now.”

They need to know that the boring, mindless elements will be taken over by the AI so that they can direct their attention to the more creative aspects of their job.  Show them what is “in it for them.”

 

More importantly, it’s up to you to support your employees if they are in the sorts of jobs which can be adequately managed by artificial intelligence.  Going back to more example of the mining industry, tell them:

 

“We’re going to (figuratively) take away your shovels because we now have machines that can dig.  What we need you to do is run, service, program, direct, and manage those machines for us.

“Since you are already familiar with our operations, you are the best people to learn these new skills and enhance our productivity.  We will offer or subsidize training to enhance your skills because we value you.”

The Takeaway

That last lesson is particularly valuable because if you don’t educate them, fearful employees will fight the implementation tooth-and-nail so that they can “keep their jobs.”  The last thing you want is active acts of sabotage by contemporary Luddites intent on destroying something they perceive to be a threat to their livelihoods.

 

We are going to need Understanding & Trust between employers and employees to make this transition successfully, in contrast to the Caution & Distrust required by the overseers of the new AIs until they are up and running in a stable, reliable manner.

 

Remember:  Use it where you have a solid plan—and a predictable ROI (Return on Investment)—but don’t forget to include customer loyalty and satisfaction as part of the ROI because that can be worth more than money!