Design of AI’s Post

🙌🙌🙌

I'm not able to go through all the comments but I'm glad to see, regardless of the funny aspect of the post, everyone agrees on the data problem we have in front of us. What leads me to be quite worry about how this is going to impact on the adoption of new technologies like GenAI or AgenticAI. At the end of the day, AI is about bringing value to the business and it's a magic circle, the more value, the more innovation we will see in this field but if we miss the magic formula of data, we will have to live again a cold AI winter.

There is a typo mistake on “artificial Intelligence”. The image fixed here:

This is so true. Today we are flooded with platforms, engines, pipelines and frameworks but what is missing is right data. The Data that is generated, reviewed, validated in the domain context. We hear this "AI will not replace you but a person who can use AI effectively will". I think this needs a change. A person with domain knowledge who can effectively use AI will not only survive but will prosper...

This is such an important topic, not only for customers but also for patients. While machine learning (ML)-based science in medicine is on the rise, very few MDs have a clear understanding about the nature of predictive analyses, modelling and simulation. However, I think this knowledge is crucial and handing over a data set to a statistician or informatician is not good enough. This Science article published by Kapoor et al. in May 2024 provides a helpful framework: https://www.science.org/doi/10.1126/sciadv.adk3452 It's complication, I know. And that's precisely why we need to learn and understand better before we apply AI and ML. As with any method, one needs to understand how it works, to avoid costly and potentially dangerous pitfalls. You don't know what you don't know. Don't be fooled by pretty, shiny graphs. Try to understand them.

It reminds of this

Thats not just today. That's always been the biggest challenge AND most important foundation. It is mind boggling to see all the hype and $$$ spent with every new shiny object put before the horses, when the majority of organizations have yet to get their horses fed correctly. It is with all things in life: whether your workout routine, ambitions to become a singer, the NBA teams eyeing the championship, parenting, etc: You MUST focus on the fundamentals, first, before you can evolve any further. Yet, almost everywhere, data management, governance, access, and data engineering, if even existing, reveal more gaps and holes than the finest Swiss cheese.

I’d add that that policy, process and culture are also key to ensuring data quality. All too often I hear clients keen to onboard chat bots or AI but if they used their incumbent non hygienic or non normalised data, the bot/AI would be hard pressed to deliver intelligent output. Oh and the graphic made me laugh out loud, so thank you 😂

I like this. That's why "quality assurance" (to use a very traditional term) is so essential to AI. Zero Trust principles in that sense also apply to AI: "Never trust, always verify". And be aware of the fact that generative AI and agentic AI don't create new knowledge, but just refactor. It is still the human that can think beyond.

Man this graphic is hilarious, love it