Why The Robots Will Run Procurement – Pub Debate Arguments (Part 2)

We gave you the highlights of last week’s pub debate here, which debated the contentious motion “This House Believes that Robots will Run (and Rule) Procurement by 2020”. Yesterday we outlined the arguments in favour of the motion, from James Marland of SAP Ariba and me. Today, the arguments against.

Jason Busch (see picture above), the founder of Spend Matters, laid into my “robot dancing” attempt, but did have some real arguments too! Firstly, he claimed that Artificial Intelligence, the central component of robotics as we define it, is not ready for real day-to-day use today (or by 2020) outside of highly targeted applications. For instance, in spend analytics, most platforms still rely on rules-based cleansing models that work. And even AI-based models still require humans behind the scenes to check on the data presented back, even once the AI is “trained”.

But even that is more advanced than AI in the purchase-to-pay area, which is even less advanced. And much of what we hear about as “robotic process automation” is from those firms who have relied on labour arbitrage in the past - if the push for AI comes from BPO firms trying to preserve margins, we’re in trouble, he claimed.

Where automation does work is in the basic, “purchasing” administrative side of the business, where much progress has been made. (That maybe contradicted his earlier point somewhat, but we’ll let it pass – there is no doubt that certain aspects of P2P have been highly automated successfully).

He also gave some good examples of where he sees that human “empathy” continuing to have real value. For instance, could a robot get to the bottom of an accounting treatment for a rebate in a supply chain finance programme (which could push it over the edge or not) in different geographies?  Or work with developing local suppliers in emerging markets, including handling tricky CSR issues maybe?

Mayank Chandla from IBM then suggested that James Marland in his speech must have been talking about this granddaughter, not his daughter, which went down well with the audience. But Chandla really focused on the timing issues again more than anything. Apparently, 60% of firms around the world still print out an e-invoice if they receive one. And only 10% of the world’s commerce is “e-enabled”, so change will come but it is relatively slow.

He conceded that AI will do a lot more and replace human activities in some areas, but this will take many years, and the ‘Bots will do repetitive tasks, assisting us rather than taking over.' As these “cognitive digital brains” (which is a great phrase) do more they will augment our decision making. But even that requires much more and more structured data for the systems to be effective. Maybe one day when that situation comes to pass we will see more ubiquitous robots – but even then, he said, humans won't ever allow the robots to dis-empower us!

I guess that last point rounds off the debate nicely. I would sum up and suggest there are three key issues here, all serious ones despite the light-hearted nature of much of the debate:

  1. Our timing was out; 2020 is too soon, but there is no doubt “the robots” (AI, machine learning, automation) will eventually take away many tasks currently carried out by procurement. It might be 2022, 2025 or 2030, but it is coming.
  1. As that happens, can procurement adapt itself to different tasks and roles, with support from the robots but proving to our organisations that our human brains can still add some real value? If we can’t do that, the profession won’t survive in anything like its current form.
  1. How much will humans allow robots to do? Will we let them choose suppliers, agree pricing, “negotiate” with suppliers? Or will we draw a line, even if the machines are theoretically able to do such work?

Thanks again to all our speakers, our chair (David Smith, pictured here with me) and audience, the nice people at The Clarence pub in Whitehall, and particularly SAP Ariba for sponsoring the event. Who knows, we may have another debate in the Autumn!

 

Discuss this:

Your email address will not be published. Required fields are marked *