AI: Whose Data Will You Use?

As you search to understand AI technology for your business, consider the data you use and a responsible plan to adopt potential solutions.

Matthew D Edwards
November 6, 2023
Artificial Intelligence (AI) visual

I've been around long enough to remember a high school Calculus teacher tell us that we couldn't use calculators until we proved we could do the work with a pencil and eraser. His argument was that if you don't understand how to get the answer with a pencil, you won't know when the calculator gives you a bad answer. That logic made sense to me and is a principle I have followed through all the years after in the form of...make sure you understand what you are committing to, saying, and doing.

Artificial intelligence (AI) is similar, and different.

AI is the same in that, if you don't know how to find the answer on your own, or aren't willing to do the work to understand, how will you know if an AI implementation gave you an incorrect answer? Do you want to bank your business reputation and solvency on that position?

AI is different in that, my TI-85 calculator only performed math with the inputs that I provided and the formula it used to answer me was fixed, i.e., a static math library. While you own the question that stimulates the response, AI uses a much larger set of inputs that you don't control called a Large Language Model (LLM).

One way to look at it might be, instead of there being a Trigonometry library, Calculus library, and Advanced Algebra library from which my TI-85 derived answers for me in a predictable, repeatable manner, an LLM is composed of libraries upon libraries of information that span all types of data it has come across to date. For example, while many people have deep, vertical knowledge in perhaps one or two bodies of knowledge, an LLM is composed of very many deep, vertical knowledge bases which may span biological, chemical, anthropological, mathematical, physical, entomological, archaeological, agricultural, historical, political, geological, agricultural, theological, philosophical, and just about anything else you can imagine.

From these very large data sets, the AI is able to see patterns, contexts, and the way language is structured leading it to suggest answers based upon what it predicts should come next.

The value of a LLM is the potential for an AI to see patterns in disparate datasets too large for the average human mind to ingest and process, and then provide potential answers in natural language that regular folks can understand.

If you're a leader in any company trying to understand what all of this AI hype is, and why you may care; or, you are interested, but don't know what you need to consider or do first, let's keep going.

Interesting conversations happen when people start wondering how their job, company, and/or personal life are going to be impacted by this technology. And it is still early in the journey to understand how to address these concerns. However, there are outstanding positive opportunities that AI may have on our jobs, companies, and lives. And, there are pretty significant risks if you don't have a plan.

For example, if you were to leverage ChatGPT in your business today, unless someone accidentally or on-purpose dumped a portion or all of your intellectual property into a ChatGPT session, when you are using this technology, it will only be drawing upon whatever publicly accessible data it has had access to date. That might be amazing when it comes to understanding your competitive landscape, generating proven sales strategies, exploring marketing messages that resonate with your target buyers, content generation, total addressable market profiling, mono- or pan-industry economic outlooks, politics, and legislation. Just remember to trust, but verify the results.

Just remember to trust, but verify the results.

However, if you want the power of AI trained to know who you are, where you've been, and help you figure out where you need to go next, you will need a different strategy. Instead of relying on someone else's data to make decisions about your specific business, you need your own, clean data; because like any other conversation in business and technology, good data in, enables good data out, and inversely.

Trility Consulting Recommendations

1. Don't be afraid of exploring artificial intelligence technologies. In fact, get busy discussing this technology internally, learning what is available and how it might positively impact your teams, business operations, and clients. This particular technology is far-reaching and isn't a fad. Businesses are being changed right now for the better. Be purposeful.

2. It is a non-negotiable requirement for you to put together a plan regarding what you will explore, how, where, and by whom it will occur, and under what conditions. In other words, you need an adoption plan that will include your data, security, governance, and associative controls.

Good luck!

– Matthew D Edwards

Read the Rest of the Series

Author's note: There are many other concepts such as natural language processing, generative AI, machine learning, artificial general intelligence. Or to qualify things a little lower, multiple algorithm/solution classes, i.e., subjects, in the Artificial Intelligence ecosystem such as rule automation, graph search/explore, constraint satisfaction, generation of statistics, data generation, clustering, and classification. This material is introductory in nature and therefore not designed to address all things, let alone all things at depth.