Why AI is good at automating tasks and struggles with automating jobs

?People should stop training radiologists now. It is just completely obvious that in five years deep learning is going to do better than radiologists.?

Professor Geoffrey Hinton, 2016 (A well known pioneer in neural networks)

Fast forward eight years, we still have radiologists. What happened?

Why you should focus on a task

Geoffrey Hinton was right about one thing. Image recognition made vast leaps, and AI software is, as we speak, as good as radiologists in detecting irregularities today. The challenge simply is that the job profile of a radiologist consists of more tasks than only "Interpreting the results from diagnostic imaging procedures to determine diagnoses". To help you understand, I curated a list of a couple of tasks a radiologist does on a daily basis and rated it by the potential of AI taking over.

Assess technical feasibility and business value to define the potential for AI

As you can see, certain tasks have a higher potential to use AI. The detection of irregularities on the scans as well as writing a report with the results and diagnosis, for example. It has been a while since my last X-Ray, so don't quote me on this. The AI does a great job in spotting irregularities; that is what it's good at. Nevertheless, you should still keep a human in the loop, since a wrong diagnosis can have severe consequences.

The handling of the scan and the procedure itself still needs human interaction, as well as informing the patient about the process and the results, in my eyes, needs a human touch and understanding.

AI empowers employees

The example of the radiologist gives you a good example of how image recognition helped to make radiologists more productive—it did not replace them. This is similar with Generative AI. I see a lot of potential in empowering employees to become more productive and automate certain repetitive tasks. Nevertheless, it empowers and helps your employees to become more productive and automate certain tasks.

Three things to remember

1. Identify the right task: Focus on tasks with high AI potential (high technical feasibility and high business value) for your business.
2. It is hard to predict the future: We saw a lot of predictions around technology. Self-driving-cars are still not mainstream for example. Therefore, focus on what already works instead of going for the holy grail of an AI that can solve all your business problems.
3. Ethical Considerations: Integrating AI technologies, particularly in sensitive domains such as healthcare, can be challenging and has a higher need for keeping a human-in-the-loop.

Hope you enjoyed today's read. It would mean the world to me if you share it with your network and friends to grow the audience.

Cheers,

Matthias