AI and the specter of unintended consequences - Workday speaks

Den Howlett Profile picture for user gonzodaddy November 15, 2017
Who will manage the unintended consequences of labor optimization that is aided by machine learning and other forms of artificial intelligence? Workday steers a narrow course.

Leighanne Levensaler
Leighanne Levensaler, Workday (via the author)

One of my current areas of focus is on the outcomes of implementing artificial intelligence (AI) solutions. This is because while organizations are motivated to use techniques like machine learning to find good answers to business problems, we cannot ignore the fact that AI solutions can and have led to unintended and, sometimes, worrisome consequences.

The unintended elephant in the room

The classic example from today is that of Facebook.

I am sure that the founders of Facebook never intended for their automated advertising system to be used in nefarious ways and yet that is what happened during the 2016 US presidential elections. The outcry among politicians and legislators rumbles on but so far, US lawmakers have stopped short of proposing laws that would act as controls over what companies can do with data and algorithms.

Going to The Hill

Some technology companies are alive to the potential risk of unwelcome and/or badly formed laws. Workday is among them, as it should be, given that it processes data on millions of workers. So it was that at Workday Rising Europe 2017, I met with Leighanne Levensaler, SVP Product Strategy at Workday to discover her view on the topic and ask what the company is doing in that context.

I am an optimist and believe that the advances and development of new work practices, accompanied and informed by AI is a net good. But we are part of an industry that hasn't really had good rules around it except to the extent that we have to be compliant across different domains and keep our customers compliant too. Even so, we are lobbying ahead of any legislation because we believe the industry as a whole should be given an opportunity to first develop codes of conduct around AI.

I was recently on Capitol Hill providing examples and I think it's important to know that there can't be a one size fits all legislative model because we're not like consumer technology. We're building technology that hopefully will improve overall operations. (see video at end)

Workday is endeavoring to get ahead of the many problems by imagining what the workplace framework will look like as the shift towards freelance work takes hold. According to Ms. Levensaler, customers are asking for several kinds of optimization:

We want to optimize our labor spend. We think it's the last frontier of optimization, meaning there's optimized customer experience, there's optimized supply chain. This is an opportunity for us, cost and quality and engagement are really important to us. Workday, can you help us? You already track people but they want us to do more. And the more comes in the form of better decisions from tools, a better marketplace, better trade-off scenarios upon which we can act, both internal and external. These are highly complex problems and I think we have a lot of potential to do something meaningful.

Dislocation by design?

My problem with this argument is that the notion of labor optimization is really an argument for dispensing with labor altogether and the dislocation that brings to both people and community.

In that context, I am not so much concerned with the fact that technology-enabled decisions and automation mean that work is displaced. That's an expected outcome. I am more concerned that the advances we are seeing coupled to the breakneck speed at which that is happening presents a genuine risk for the fabric of society as a whole.

Ms. Levensaler doesn't disagree but says:

When I first got out of school, I was an organizational development specialist. And part of my role was to think about how the internet was going change jobs and skills. So it was almost like today is yesterday. We thought of many things but totally missed SEO and you know that's become a huge business. I read somewhere it is worth $80 billion. And we missed that. Who is to say that we won't find fresh uses and entirely new industry segments as a result of what algorithms provide as answers to today's problems?

That is a fair point but it is not inevitable that enough 'new' emerges from the current disruptions to balance the books - so to speak, or that it will occur in time. The problem, as everyone recognizes, but which has everyone a tad stumped, IS what happens in those scenarios.

The fact Workday has turned up on Capitol Hill and is making the case for business algorithms and systems to be considered separately from the consumer Internet is a good thing. But is the work that companies like Workday are doing going to help make the world a better place for the humans they help manage or will the optimization algorithms serve to undermine the workplace?

Ms. Levensaler believes remaining close to customers is critical.

It would be wrong for us to pretend we have all the answers and so for us, problem-solving for our customers that takes them a bit further forward is how we should proceed and that is what we're doing.

My take

I admire Ms. Levensaler's forthright approach to both the known and, as yet unknown, technology impacts. The fact Workday is making sure it is out front with legislators and standards setters is all to the good as a necessary first step.

I am yet to be convinced that Workday (and every other vendor) has come to a place where they weigh the alternatives as a method of discovering the right paths. Ms. Levensaler, for instance, while acknowledging a bumpy road, is informed by her belief in an optimistic view of the future. That is perfectly valid. But what if that's wrong?

For example, as I was rounding out this story, an email came in drawing the analogy between current impacts with the introduction of ATMs. The argument went that ATMs displaced many bank workers but that the resulting cost savings allowed for an expansion of bank networks and the resultant increase in bank branches, including a fresh influx of staffing needs. That was true - until the call center emerged and then bank branch networks shrank in favor of low-cost service operations. As a side effect, customer service levels fell off the proverbial cliff and, I would argue, have never fully returned among some banking groups.

Will we see the same as labor groups are optimized and if so, what should be technology leaders' position?

To be continued...

A grey colored placeholder image