In the Functional ExtAInction Battle white paper, the authors repeat the ridiculous claim that we may see AGI within the next decade (despite their acknowledgement that it will need a gargantuan data processing infrastructure which will require an unachievable energy requirement that is the equivalent of 34 new nuclear power stations by 2030 and 34 more by 2035, which still overlooks the requirement for an exponential increase in training data, which does not and will not exist — as we don’t even have enough data today for effective LLMs, which already stole every piece of data on the internet to get to the point where it fails almost 50% of the time on tasks it was specifically trained for, like ChatGPT which generates responses with incorrect information 52% of the time! [And now that the internet is filled with as much AI slop as actual human created content, performance is just getting worse.])

The reality is that we won’t. Every 10 years we see a resurgence in AI hype and every 20 years it is a big one. This is the biggest AI craze since the early 80s, when Japan almost bet its entire IT economy on 5GL which was supposed to allow a computer to solve a problem presented to it from constraints alone. i.e. A first attempt at … AGI! It didn’t happen them, it’s not going to happen now. Until we crack intelligence and ingenuity, and how to effectively model it, all the processing power (and flawed data) in the world is not going to do that for us. Moreover, we will continue to outsmart the most advanced AIs in the world using toddler level thinking and hiding in boxes and bushes, like these marines did.

However, the C-Suite’s desire to believe in this is forcing digitization efforts, if not AI, and the teams that survive this current nightmare will be those that not only digitize under the guise of AI (since every vendor claims it, whether they really have it or not and since most C-Suites have no clue what AI is or isn’t), but in fact embrace next generation automation on steroids (like Adaptive RPA that learns from every exception, decision, and override to continue to decrease the need for human intervention over time). (The teams that fail to embrace modern tech will be sidelined and the teams that don’t resist the experimental AI being pushed on them by the overpriced consultancies brought in by the C-Suite will continue to contribute to the 95% failure statistic and possibly end their function entirely.)

Moreover, the teams that embrace appropriate digitization, like form fit ARPA, will evolve into the evolutionary niche that the business needs, and that only Human Powered Procurement can do. While the authors got the premise wrong (it’s not AI that you need to worry about, it is the AI Marketing), and are overzealous about the emergence of AGI, they get the future right. (Which is the past, by the way, but more on this later!)

According to the white paper authors, the future is spend that needs human interaction (because it can change the game or presents existential risk to the business). This is what should be Procurement’s primary purpose. Procurement’s never had enough time to manage all spend, and like the authors note, it shouldn’t be managing tactical spend. It should be automating it. If the spend is low volume, low risk, easily replaceable, etc., Procurement should define the best-practice strategies and processes and let modern tech (which doesn’t require AI*, by the way) entirely automate it. This goes for routine and non-routine spend where in the case of non-routine spend, the provider has best practice templates culled from its experience and community intelligence. However, routine strategic spend will not be turned over to AI. It will be highly automated, but human experts will still vet the suppliers and verify the decisions before a contract is signed or a PO is sent out, but a lot less time will be spent on strategic spend that is routine and usually doesn’t change much from year to year.

The world, and the technological underpinnings, will continue to evolve as they have for the last four decades. The pace will pick up a bit, but not much. Humans are naturally lazy and change resistant, which means that significant change typically requires a generation (or two). It’s never a “whole new world”, just a slightly different one. The only time humanity has ever undergone and grudgingly accepted such significant change is as the result of a significant natural or man-made disaster that has devastated entire cities and populations, and forced adaptation to survive. But rarely has the survival brought something better in the lifetime of those forced to undergo it! (Plus, the world has been a commercial hub since before history was recorded. We’ve always traded to survive, thrive, and satisfy our desires. It’s just that we’ve replaced food and trinkets with digital bits that represent food and trinkets with a digital equivalent of their perceived monetary value. So whether you call the function in the business that manages that aspect of the outside world Procurement or the Commercial Hub [of the business] is irrelevant.)

So don’t fear a rapid change, it’s not going to happen. But prepare for a steady change, and you can keep up while your peers fall behind.

^* When the doctor wrote his AI In Procurement (Sourcing, Supplier Management, etc.) Today, Tomorrow, and the Day After in the late 2010s (before all the Gen-AI bullcr@p), what he didn’t tell you was that everything he included in “the day after”, which is the majority of everything the Agentic AI providers are promising now, was already possible. It just required a lot more code, sweat, and tears on yesteryear’s stacks with a lot more templates and customized training data sets than most providers, or companies, had at their disposal at the time.