Why understanding how Gen AI works matters

Generative AI, or Gen AI, refers to the tools you likely use every day now.  ChatGPT, Claude, Copilot, Perplexity, Grok, and many others.  For most of us in our day-to-day lives, it's used to get information, take a picture of something, identify it, or find somewhere to buy it.

Gen AI is actually somewhat broader than the ChatBot we all use; think of it more as an application/product that uses generative AI models. But, jeez, I am getting down into the technical weeds already.  

So, we can accept that a tool like ChatGPT is built on top of Large Language Models (LLMs). The Model is what enables ChatGPT to return the results we all get when using it.  This is important, as I am going to take us all on a journey from how AI gets and processes data to eventually creating a model (the LLM) that then provides us with information.

For me, this is really important. It's one thing to just type something into ChatGPT or some other AI application, but it's equally important to understand what happens behind the scenes.

Dependancy

Love AI, hate AI, or anywhere in between, many of us are becoming dependent on it.  AI has so many amazing uses.  Are we becoming so used to it that we are becoming too dependent on it?  That's a question I will be asking a lot on this site.

But, it's so good

I speak for myself here.  I use it every day for so many things, both personally and at work.  Some examples.

Personal

I admit that it has become such a significant part of my life.

• Coding: This site was built entirely with AI.  I can code, but an important part of this site's existence was handing the coding over to AI to show it off.
• Research.  All sorts of research.  I watch or read something, and I think, "ok, what does that mean?" or "I want to find out more."  What do I do?  Ask AI.
• Take a picture of something and ask AI, what is this? How do I do something with this?  
• Workout routines: I use AI to mix up my weightlifting sessions. 

Work

Even at work, I use it for a range of tasks, encouraged by my employer, who has invested a lot of time and money in getting the best out of it.  Yup, we know where that could end up.

• Research: I use it all the time to find information and check data, rules, regulations, and more.
• I use it to sort my mailbox; it reads my mail and prioritizes it.
• Checks my notes and provides an opinion on where they could be stronger.

I could go on, but you get the idea, the uses are significant.

"I never use AI"

There are many people who say this.  I believe them.  I know a lot of people who refuse to embrace AI, whether they don't believe in it, fear it, or for whatever reason. I am not going to criticize anyone's views and tell them what they should or shouldn't do.

I would say AI is here to stay.  A belief I do have is that those who embrace it and those who don't will be in quite different positions, and may already be.

For example:

I work in credit risk, and that job requires a lot of reading, thinking, research, and decision-making.  Each case I work on used to take 2-3 hours.  It now takes me just over an hour.

For me, Copilot is the tool provided.  To be clear, I don't outsource everything to AI.  The way I see it, I have a personal assistant, Copilot.  It's there to handle tasks that take me a lot of time but add little value to the process, so I can benefit from not having to do them.

I still do the following

• Review everything
• Do the thinking
• Make the decisions

What I don't do anymore

• Read every document line by line
• Spend time on the internet finding information
• Manually reviewing policy documents
• Checking calculations

I have been doing my job for a long time, 35 years in various formats and companies.  I understand that AI output should never be implicitly trusted.  It's a balance: saving the right time, achieving the same result, or, admittedly, better results.

My point is that I can create the same output and quality in half the normal time without AI.  From an employer's perspective, who do you want?  The person using AI doing a job in 90 minutes or the person who isn't using AI taking 180 minutes?

I'm making a point, oversimplifying it, but it's the reality.

The danger of becoming the assistant

Something I am wary of is the danger of the role being switched.  Instead of AI being the assistant to the person, the person becomes the assistant to AI.  How would that happen?

Instead of the person giving AI the right tasks for the right reasons.  The person gives AI everything needed to complete the tasks, and AI asks what it needs to complete them, does the thinking, and makes the decision.

For me, I recognize risks and prevent them from happening.  Will everyone else?  Will companies ensure it doesn't happen?  

Losing the ability to think

This is something else that worries me.  If we offload too much to AI, will we stop thinking, or will we not know when something is going wrong, or when a decision has not considered the right information, simply because thinking has become significantly less important as our reliance on AI has lost the right balance?

For someone like me, with 38 years of work experience and AI not being a thing for 36 years, I am not concerned; I believe I understand AI enough to control it, not let it control me.

Where I do get concerned is with young people who are already using AI to write essays, get all their advice from it, and generally accept AI as right, even when it can be convincingly wrong.  It's an important consideration.

Let's move on

AI, it's great, it's bad, it's interesting, it's dangerous, and anything else you want to add.  I intend to get into all of that once we understand how your AI App works.

Next, I am going to look at how AI gets all of its information.  You can read it here.