Word of the year?

 People are talking a lot about Artificial intelligence - so much, it will probably be 2023's "word of the year". But when talking about AI we usually focus on the algorithms and the outcomes - and don't get me wrong, they are usually amazing in their own ways! But let me whisper you about two dimensions that are usually skipped:

- the first thing is about data. AI runs on data, so the kind and quality of the data you choose to input in the model (whether it is a Marketing CRM, a Chatbot to build trust and superiority on your product or a HR Attrition Predictor,...) will have a great influence on the outcomes from AI. The cleverer the data, the better the outcomes. And data choosing is typically a human input in the model;

- the other thing is that most of AI tools give you predictions. That's their output. But the job is typically not finished there - predictions by themselves are not the end game. It's what you do with them. It's the actions that you (a human) take from those predictions. The predictions enable a better decision making - but the calls are typically still done by you (even if you are incorporating them in a Communication Ecosystem built on a Customer Journey, as an example).

If you are using AI tools, don't underestimate the role of data and decision making. And don't underestimate the importance of human decisions to achieve better results.

Data vs AI

Data vs Algorithm

This is a really exquisite question – in AI (and given resources are limited in the real world) should a company invest in better data or better algorithms? Good question, right?

There is actually a study done by a Microsoft team that looked exactly at this topic – and the conclusion was… that usually (not always), collecting better data ends up being more valuable than investing in better algorithms => it generates better outcomes. The reasoning for this is modern algorithms’ performance differences are relatively small – and smaller than the difference in performance of the same algorithm with more/less data.

So… Data > Algorithm.

Now, a word of caution. There are businesses, where fortunes are made on infinitesimal differences (think about financial trading) – nobody will want to play with a slightly worse algorithm than its competitors…