Digital transformation and cyber security ought to be bigger priorities for most companies.
“How will AI change the CFO’s job?” My boss, the CEO, keeps asking me, “what’s our AI strategy?” What should I tell her?”
I keep getting these questions from students in our CFO program. To investigate ground reality, I chatted with a few CFOs and COOs, off the record. Here is what I found out.
Get your data in order:
And that is putting it mildly. AI makes sense only if the business has an excellent data warehouse. A good data management system is just the beginning. In virtually all the cases I encountered, in the relatively smaller larger cap, mid cap and small cap businesses, companies continue to struggle with legacy systems that are inefficient and have outlived their effective use a long time back. Chronic under-investment in updating current information systems is a pre-requisite to get anywhere close to harnessing AI. Mid to senior level executives continue to struggle with Excel spreadsheets, data integrity issues and employee awareness and competence with data. Part of the problem is that many employees are not analytical enough to be able to use data to assist in decision-making. A data and accounting system literate workforce is a pre-requisite for running the business well. AI is much further down the road in the life cycle of these companies.
Cyber risk is a bigger priority:
A cyber attack is an inevitability, like getting Covid. It’s not a matter of if, but when? And many businesses are simply not equipped to withstand such an attack. Contingency plans, if such an attack were to happen, are generally half-baked. Cyber response drills are carried out in patches but not in a coordinated way because a thousand meetings and short-term firefighting get in the way. I have heard of businesses where employees of consumer facing businesses are still writing down credit card numbers on a spreadsheet. It will take one disgruntled employee to let in a cyber worm, intentionally or otherwise to wreak havoc. Cyber security is hanging by a thread in many firms. Before thinking about AI and other such new shiny objects, work about fortifying your cyber threats.
An executive rightly points to the role of the firm’s culture of investing: “cyber security and cyber risk probably all goes back to culture. We need strong investment in technology and good detailed standards.” Investing in the foundational blocks is crucial, even if it’s invisible, like cyber security, which I suspect for non-technology companies is often reactionary.”
Everything is called AI now:
Simple decision trees, say linking symptoms to a disease in a hospital, is now relabeled as AI. Many chat bots are pitched as AI but underlying this are simply data instruction flow charts. A senior executive of a midcap company says, “I’d probably say that what I’ve seen is Data Science ostensibly called AI at this point. I’ve certainly seen a lot of AI / ML (machine learning) — and sometimes it almost feels like almost any advanced technology is labeled AI.”
This is not to say that AI cannot be transformational in certain contexts:
AI is certainly transformational in certain companies and that list will grow in the future. If you are a Google or an Amazon, that has solved its data architecture and cyber threats for the mast part, AI can multiply your worker productivity manifold. For the most part, AI is mostly the realm of power players and very large cap digitally evolved companies. Of course, there are many niche settings where AI can be revolutionary. A top scientist I know reports that the time it takes to develop drugs has easily fallen by 50%, if not more, as AI can easily run combinations of plausible molecular chemicals with the greatest chance of success that scientists can investigate. However, most of us run the risk of over extrapolating those niche situations to our businesses.
A senior executive of a midcap company said to me, “I suspect a lot of the document related AI is quite good. I also suspect that AI would be quite good going over unstructured data and deriving insights (e.g. customer support emails) — and again, it just needs access to that. Text can be leveraged off the corpus of the Internet as opposed to very domain specific text — but this, of course, depends a bit on good prompts. I think the CEO of Roblox at the Wall Street Journal (WSJ) Tech Live conference was talking about how they’re using AI to reduce staffing around trust & safety — likely running sentiment analysis over chats, etc. — which I totally believe as a viable use case (and outside of the data warehouse.) Anyway, I guess my point here is that depending on the AI strategy and the goals, there are certainly targeted areas of AI that are worth exploring / a company can explore even with a messy data environment.”
My comeback to this: is it worth managerial time and effort to invest in niche AI applications when your data organization and cyber risks are screaming for attention?
So, what should I do?
For most CEOs, AI is a PR event in most medium and small companies. Get your data house and cyber risk in order first.
