Counting Pennies.
This was written by a human. :)
Last week The Economist stuck a noteworthy pin in the artificial intelligence balloon.
The article What happened to the artificial-intelligence revolution? noted that the technology has thus far brought no recognizable economic value to its users.
Makers of AI? As we all know, a different story. Roughly $2 trillion has been added to the market value of Alphabet, Amazon, Apple, Meta, and Microsoft, a flood of venture capital ($142.3B in 2023) has washed into AI start-ups, and owners of Nvidia stock should have few complaints.
The Economist listed four reasons why the value needle has not – or not yet – moved for the enterprise: widespread concerns about data security, bias, and the accuracy of gen-AI output; widespread “pilotitis,” seen in a plague of small-scale, tech-centric Potemkin pilots – the equivalent of technology motion without business progress; small, limited scale implementations, most often in marketing and customer service, few of which will move the P&L.
Add in a corporate hesitation that often accompanies rapid innovation; when, leaders ask, will the carousel of artificial intelligence slow enough so that we might jump on and invest in something that won’t be obsolete in six months?
All told, the rate of enterprise adoption is far slower – and with much less economic impact -- than McKinsey’s “The stage of AI in early 2024” report and Microsoft’s “2024 Work Trend Index Annual Report” might lead us to believe. Indeed, The Economist article reported that the Goldman Sachs’ stock market index of firms that have “the largest estimated potential change to baseline earnings from AI adoption via increased productivity,” shows that the share prices of the high-potential firms have, since OpenAI’s ChatGPT launch in Q4 022, failed to outperform the broader market.
In other words, investors have seen no prospect of extra profits.
No prospect of extra profits.
To which I would say . . .
Not yet.
Time to take a deep breath and adjust expectations.
First: artificial intelligence is not a silver business bullet.
Yes, AI is amazing and it’s breakthrough and it is an inflection point – all true – but it is a set of tools that must be carefully integrated into multiple business processes to fulfill its promise. This will not come quickly nor simply.
As colleagues Milan Turk, Jr. and Don Bassell point out, the use of AI within a business will be a penny-in-every-pot play, one in which value creation will be a step-by-step accumulation of basis points throughout a business process. Over time, the value will grow, even multiply.
Thanks to the use of one or more AI tools, a process will be faster here and smarter there; it will deliver new insights and better decision options; it will help us know sooner and deeper.
It will re-shape not only this process, but that process. And that one. And that one.
In return: a penny here, a penny there; pennies, soon, everywhere.
The AI-in-Retail Research Group is now evaluating the potential, quantifiable value of artificial intelligence as applied in 141 standard retail industry workflows across eleven core business competencies. Our initial assessment would surprise few industry veterans: the application of AI to one single workflow will make no more than a small value difference, one barely felt by a line-item on the income statement.
However, the application of AI across a growing portion of the entire competency (for instance, store-level hiring and training) will, over time, create a significant, move-the-Goldman-index difference.
How much time?
Goldman Sachs has, with a broad brush, pointed to a measurable impact by 2027.
For retailers, I’d say it depends.
I am aware of leading retailers who are collecting more than a few pennies right now. (In fact, several are multiplying pennies -- most often quietly, so that competitors are not tipped off.) I am aware of leading retailers who are now wisely preparing – through data preparation groundwork and strategic reviews – to begin collecting pennies late this year or early next.
But it depends.
It depends upon a brand’s willingness to invest in data quality and accessibility – knowing that AI is the refinery, and data is the oil. Internal and external data. Structured and unstructured data.
It depends upon a brand’s willingness to devote not only IT but business unit resources to the task, knowing that AI’s optimal value will not result from one-off solutions, but from workflow transformation. And that AI-enabled workflow transformation – the augmentation, acceleration and disruption of a business process, step by painful step -- will require expertise from BU leaders, operations, and HR as well as data architects, system architects, integrators, and UX designers.
It depends upon properly asserted expectations at senior levels. The conventional wisdom suggests that any all-stores technology roll-out will take 18-24 months from ideation to production implementation. The effective use of artificial intelligence tools in any one corner of the company – if pursued at an improve-the-business, not a Potemkin pilot level – will demand no less.
It depends upon a brand’s culture, its persistence and constancy to get stuff done – knowing, always, that the path will never be straight, regardless of the consultant’s marching chevrons slide. Kneejerk changes in leadership (and thus, direction and investment) will kneecap progress.
Message to London: for the enterprise, the AI revolution’s just begun.
I am Jon Stine, 35+ years in retail business and technology. Most recently in conversational AI.
I read, I listen, I observe. I think, I write, I advise.
https://www.j.christopherccv.com
+1 503 449 4628.I