What I learnt from this year’s AI conferences
Dear IT Manager, Data Scientists, Product Owner, CxO,
I sincerely hope the words “AI” and “machine learning” have become part of your daily vocabulary by now…
Either way, it’s time we have ‘the talk’.
To no fault of their own, I’ve simply seen too many talented people fail at their AI projects. That ends today.
Based on how we’ve approached & successfully delivered AI projects at In The Pocket + my learnings from 2018’s AI events I attended, here are my key takeaways to set you up for success in 2019.
When I was at ML Conference in Berlin last week, I had an interesting conversation with one of SAP’s AI leads. On what to her were the 3 most common reasons why machine learning projects fail.
According to her, you will fail when:
- your skilled ‘team’ is simply a bunch of people;
- you don’t properly prepare your data;
- you don’t succeed at meaningfully translating machine learning models into a production environment.
I heard the same message from Google’s Chief Decision Engineer, Cassie Kozyrkov, at World Summit AI earlier this year.
What I’ve learned is that it doesn’t matter how incredibly well the farmer cares for his crops; if there’s no researches to create ovens; if there’s no-one who designs thoughtful recipes; if there’s no chef capable of translating these recipes into culinary delights.
Let’s try that again:
It doesn’t matter how incredibly well your data engineer aggregates & prepares the data; if there’s no-one who creates useful algorithms; if there’s no-one who designs thoughtful machine learning models; if there’s no-one capable of translating these to impactful digital products.
And more importantly, your AI project will fail if your team doesn’t have a proven track-record of working together in the kitchen & actually serving up dishes that aren’t cold, tasteless or even horrible to eat.
That’s why at In The Pocket all of our teams are multidisciplinary. Everyone in their team sits next to a Solution Architect, next to a Designer, Tester, Machine Learning Engineer, Software Engineer, Competence Lead, Product Manager, and next to a Strategist.
Moreover, Software Engineers and Product Managers have had extensive Machine Learning training. As they obtained these ‘fusion skills’, they now form the bridge between different disciplines.
This way — as we match business cases with user needs & create smart interfaces — we collectively think about scalable and secure architectures, useful data, code quality and task success, from day one.
We don’t just think of creating abstract machine learning models or isolated proof-of-concepts. We leverage our years of experience to help you successfully apply AI to your business process & tools.
I recently gave a talk on innovation before one of Belgium’s leading telcos innovation teams & got asked this question: How do we stay ahead of the (S-)curve? How do we make sure we stay up to date and successfully apply the latest technological trends?
Simply hiring PhDs and hoping they can cook, is hoping for the best — and hope isn’t a good strategy. Decisions at scale (which Machine Learning is) needs skilled teams. Diverse teams, too. Our AI team ranges from people with a PhD to engineers with a background in music.
As it’s hard to read the label from inside the bottle, that was still a very valid question though. The answer is however easier than you might think:
You have to bring in new perspective. Bring in external teams that have only one mission in life: to absorb the latest technologies & apply them to useful business cases to solve real user needs.
Bring in In The Pocket.