I developed AI at IBM. Here's how to not become intellectually dependent on tools.

• Sol Rashidi has worked in AI for 15 years, scaling capabilities at companies like IBM.
• She said workers have to continue to use their brains to avoid intellectual dependency on AI tools.
• Rashidi uses it for acceleration, not replacement, and advises against copy and pasting responses.

This as-told-to essay is based on a conversation with Sol Rashidi, a former tech executive at IBM, AWS, and Estée Lauder, who is based in Miami. The following has been edited for length and clarity.

In the last 15 years, I have built and scaled AI capabilities, and I have over 200 deployments under my belt.

I went from being an individual practitioner to running IBM's enterprise data management practice. I was the chief data officer at Sony Music, the chief analytics officer at Estée Lauder, and the head of technology for AWS's startup division in North America.

All my experiences from 2011 on have led me to realize there's a real chance people will develop a codependency on AI. So I'm focused on workforce preparation and educating the masses.

Now I have my own company where I'm working on solving the problem of AI in the workforce by teaching enterprises how to prepare their workforce for the future, and how to use AI and automation to amplify the workforce instead of eliminating it. If you're going to use AI in your day-to-day, great — but you have to be conscientious, to outsource tasks and not your critical thinking. You need to avoid intellectual atrophy.

Intellectual atrophy is when you lose your cognitive ability to think critically because you're outsourcing that thinking to tech. Just like our muscles atrophy if we don't use them, so does our brain. The big thing that you've got to be careful of is making sure that generative AI doesn't make your thinking become generic, because everyone else is also using ChatGPT. You maintain your edge by using cognitive power.

Don't replace your work

As an individual, I use six to eight AI tools every day. I use AI a lot for data processing, so I can think about the patterns and insights and, from there, observe and spotlight frameworks and models.

But when using the tools, I always ask myself, "Am I using this to accelerate work I have to do, or am I using it to do the work for me?" It needs to accelerate the work so that the thinking is left to me. "This is making me faster, but is it making me more capable? This is making me more productive, but is this making me more valuable?" I use the tools to expedite and facilitate versus doing the work for me.

Part of what I do is communication, and I don't ever want to lose that edge. I don't use AI to write emails, keynotes, or personal interactions.

It's really important for me to be able to understand whether or not what I'm communicating is being perceived in the way I intended. That takes practice. Anything that comes from the heart or mind has to be sincere, expressive, and communicate the right messaging. It has to be organically generated by me — no exceptions.

Don't copy and paste

We live in a society right now that values convenience over competition and speed over substance. But the key to keeping up is actually slowing down, because there is no shortage of information coming to us. We're ingesting so many gigabytes of data every day through WhatsApp, Slack, email, LinkedIn, and Instagram. The way we used to handle the workload of the past cannot be replicated to handle the speed of today.

So we have to develop our discernment muscles, which is the ability to spot a signal from noise. A large percentage of content worldwide right now is AI-generated, and we have AI-generated content that is being cannibalized to retrain itself.

Moving forward, we're going to get to the point of diminishing return. Problem-solving skills are going to be so important, and it will be super important to discern, validate, and verify AI responses. You can use AI to author the first draft, but maybe don't copy and paste the output because it's often inaccurate. Think of it as a first draft always.

The last team that I managed was a data science team at a Fortune 500 company. I tasked my junior and senior data scientists to come up with an approach for the CMO for a new product. My junior scientist produced the same deliverable as the senior scientists but it took them less than half the time because they took the word of ChatGPT.

It sounded great, but they short-circuited the process of research and verification, so I had to make a new mandate that they cannot use AI to do the work for them, but only to help facilitate and accelerate the research.

I told my junior scientists and anyone highly codependent on AI, "I'm paying for your brain and uniqueness. I'm not paying you to copy and paste, because, quite frankly, a license for enterprise API from OpenAI is a lot cheaper than you."

It's so easy to ask ChatGPT a question and get an answer that sounds really good. But if you don't use critical thinking and depend on yourself to solve problems, you could be outdated within a few years.

How is AI affecting your work? Contact the reporter via email at aaltchek@insider.com or through the secure-messaging app Signal at aalt.19.