This article is more than 1 year old
AI in the Enterprise: How can we make analytics and stats sound less scary? Let's call it AI!
New names for old recipes
Register Debate Welcome back to the inaugural Register Debate in which we pitch our writers against each other on contentious topics in IT and enterprise tech, and you – the reader – decide the winning side. The format is simple: a motion was proposed, for and against arguments were published on Monday, another round of arguments today, and a concluding piece on Friday summarizing the brouhaha and the best reader comments.
During the week you can cast your vote using the embedded poll, choosing whether you're in favor or against the motion. The final score will be announced on Friday, revealing whether the for or against argument was most popular. It's up to our writers to convince you to vote for their side.
For our first debate, the motion is: Artificial intelligence in the enterprise is just yesterday's dumb algorithms rebranded as AI.
And now, arguing FOR the motion is RUPERT GOODWINS...
Algorithm is a lovely word. It sounds technical and mysterious, hinting at its origins in the exotic mysteries of 9th century Persia, but it just means recipe.
The use of the exotic to mask the mundane is indeed diagnostic of what’s going on. Let me reveal the master algorithm underpinning all AI in the enterprise. Take a technology few in business fully understand, define it as needing lots of expensive everythings - hardware, software, people, bought or hired - and don’t say exactly what it’s doing or how it’s doing it.
Then sell it as the most important new thing any company can have, to fend off all the competition already adopting the new shiny to reduce you to rubble. Don’t quite know what’s going on? Perfect, you won’t be able to tell whether it’s working or not. Sign here.
That recipe has worked since computers were the size of warehouses, and it’s still working perfectly well in its brand new AI hat. In the past it’s been 4th Generation or Knowledge Engineering - crucially, then as now, with the vague promise of ‘proper’ AI, machines that think for themselves, but they’re still as dumb as a box of sand.
Outside the enterprise, things are a bit more interesting. Very specific tasks based on machine learning, although not particularly novel, are newly practicable now Moore’s Law has with its dying breath bestowed us transistors in uncountable trillions. Vision and speech recognition can replicate the processing pipelines of human perception - algorithms many millions of years old - and the nagging suspicion that natural language processing is getting good at semiotics won’t go away. But unless your enterprise is based on automatic facial recognition or being unbeatable at board games, this isn’t the HAL 9000 moment you were promised.
Proper, useful enterprise ‘AI’ is analytics. It’s stats. You’ve got lots of data, and you need to know what it’s telling you. You can put it through various transforms, you can stick it in a spreadsheet, you can pump it into a neural network, you can sort it and refine it in a database, but you’re doing the same basic task that relies on three rules: your data must be good, your tools must be your servants not your masters, and you must be able to understand and test the answers.
That’s not a problem exclusive to AI: AI’s problem (to you) and promise (to the marketeers) is that it sounds less scary and more exciting than analytics or statistics. It promises to automate away your ignorance, your discomfort, and to give nice solid answers to questions you suspect are important but which you can’t quite frame. Yet in the end the enterprise’s survival and prosperity in the market depends on deciding what data matters and how to accurately collect it, and AI can’t make those decisions for you. You need to properly analyse what you find, and then to formulate and act on a business model powered by that data. If AI is taking away your knowledge and control of that process, it’s putting you in danger.
Smart people know that. That’s why in the most automated, high-powered, cutting-edge pure data players, the financial market marauders, prize their quants so highly. High speed trading has to be automated and it has to be based on the very best analytics, but without a mutant from a maths department in charge of the crunching who knows what’s going on, it isn’t going to work. AI is scale, AI is speed, but AI is the oven, AI is not the chef. The recipes remain.
In the end, it’s a question of semantics - where you draw the lines between process and algorithm. The techniques we call AI are disparate and constantly evolving, and like all evolution when you get close in it is constant change that slows and speeds somewhere on the line between static equilibrium and wholesale reinvention. Of course AI is built on existing ideas. Of course it innovates.
The important thing in enterprise IT is that you understand it. That you know what questions you’re asking, that you comprehend the answers it’s giving, and that if it’s going wrong you can fix it or throw it away. That’s the algorithm you need to implement. Natural intelligence makes all the difference, an old recipe but a good one. ®
Cast your vote and track the debate progress here.
Biting the hand that feeds IT
