Consulting in the AI era

There are times in ones career where things don't click anymore, and we lose sight of what was exciting for us in the first place. For me, realizing I didn't work at all in the field I studied (and that really excited my imagination back in the days) was a trigger to start a journey to understand and learn once again what I learnt back in the (not-so-far-away) days of 2015.

However, the field of AI is dense, and things have evolved a lot in 11 years' time. Where could I start?

Some years ago, I stumbled on a book called "Deep learning for coders with fastai & PyTorch", written by Jeremy Howard & Sylvain Gugger (who once taught maths to the very mediocre student I was), and bought it on the promise I would do something with it, which of course, was a false promise, until 2 months ago.

2 months ago, I started to take a new turn in my career, without going awol : I would offer technical consulting on AI matters for companies. Huge stuff, I know - isn't everybody selling AI in some ways right now? What could I offer that others wouldn't?

Well, as it happens, I'm a managing consultant in a firm, and used to do management / business consulting for various industries, and let me tell you this (you, being probably the only being reading me, if you're not a robot) - start-uppers, VC heroes, business angels, tec prophets all sell AI as the wonders of our days and age.

In a sense, that's true : we've seen AI helping us accomplish leaps in technology (see what the defence industry does, from Anduril to target software used by various countries in recent wars). In a sense, that's false : has AI really been implemented worldwide? Are all business processes automated? Are there even data to work on?

The answer is : it's true and false, very niche deeply advanced software & use cases coexists with basic, excel bullshit.

So why is it? I don't have any other answer than my experience & industry knowledge, but it's safe to assume a few assumptions : companies are conservative, implementation of cutting edge software requires a company to be culturally, organizationaly, architecturaly ready to drive it, with demonstrated and sound returns on investment & limited risks.

Which is not the case in many industries (a great opportunity for a latter blog!)

So what can we, gentlemen, bring to the table ?

It turns out that people making cutting-edge AI do not really understand companies business mind-set. It turn out that those who does, often sell bullshit.

Here's my spot : having the (real) knowledge of how AI works, paired with how businesses works so that I can help them implement AI where it matters, and where it will work.

Follow me on this blog on my journey. I promise, it will get interesting at some point!

Antoine