WHERE IS BUSINESS VALUE HIDING IN AI?

By W H Inmon

Everywhere you read, every news show you watch, every ommercial you see there is this overwhelming fascination and fixation with AI. The AI vendors have sold the world on the idea that AI is the future of mankind.

In a word, people and the world are stoked.

NEW POSSIBILITIES

And what is it about AI that gets people so excited? There are many facets of AI that appeal to people. AI can talk with you. AI can create presentations. AI can answer questions for you. AI can look things up and find a multitude of facts – both obscure and contemporary.

The interface, the capabilities, the range of possibilities of AI are just staggering. And people have never seen this before.

AI has opened up a whole new world.

Now, people that have never touched a computer before can do sophisticated analysis on the computer.

SELLING AI

And who has AI been sold to? AI has been sold at the highest levels of society – to the board of directors, to politicians, to decision makers everywhere. Unlike previous technologies that have been sold to technicians, AI has been sold at an entirely different level.

HYPE

And what expectations have been promulgated by the AI vendors? The AI vendors promise that AI can do anything. Like all technology vendors, the AI vendors have created a huge amount of hype – just like their predecessors that preceded them.

The promise is that AI makes everything easy. The impression that AI vendors give to the businessman is this – “just bring in AI and money will just start to flow into your lap.”

This is a very broad and powerful promise. The AI vendor has set the expectations for AI at a very high level. The hype level has been exceedingly high.

REALITY

But what is the reality of bringing AI into the corporation? In particular, what has been the experience of corporations with the reality of AI and business value? Has AI really unlocked the doors of business value flowing into the corporation?

WHAT PEOPLE ARE SAYING

The Wall Street Journal and MIT University are two respected institutions. The information provided by these institutions is relied on around the world.

WSJ says that it is impossible to measure the impact of AI on the corporation. And MIT says that even though there has been big investment in AI, there have been no appreciable or measurable impacts of AI on the corporation in terms of enhancing business value.

Where then is the business value that has been promised by the AI vendor? Is business value and AI just a mirage?

Why does AI not provide the wonderful returns that have been hyped by the AI vendors? What’s going on here?

It turns out that this is a simple question with a complex answer. There are many factors that conspire against getting business value the way the AI vendors have portrayed AI in the corporation.

MAKING THE PROMISE COME TRUE

In truth there is great business value potential in AI. But using the approaches promulgated by the AI vendor is not the way to proceed.

CONVENIENCE DATA

The first factor conspiring against AI delivering business value is that the kind of information that AI yields can be called “convenience” information.

Ai is wonderful at delivering answers. It is just that the answers that are given have little or no impact on business decisions. The answers are convenient to have answered by the person making the query. But the answers are not relevant to the issues that bring business value.

Consider some typical uses of AI –

Is my plane late?

Which metro should I take?

What was the name of Alexander Hamilton’s mistress?

What is IBM’s stock price?

Who won the election last night?

These questions are all interesting questions. These questions are all important to the person asking the question. But – for the most part – these questions have little or no value to the making of a business decision. They are tangential to the business of the corporation, at best.

Answering these question is a convenience to the person asking the question.

On the other hand, business data and the questions being asked about business data do have a direct impact on the making of business decisions.

Convenience data is very broadly focused. Business data is focused on the business itself.

So there is a big difference between the kinds of data and the questions being asked AI of the data and the questions being asked to find business value.

And because of the fundamental data and questions being asked, the business value to be found on AI is not to be found in the unfocussed data found in the public domain.

The data that can be found in the public AI environment is –

Interesting

Useful for many purposes

But not particularly useful to find business value.

The convenience data found in the public AI environment is cool to look at and use but does not lead to more profitability or to larger market share.

IS THERE A COMPETITIVE ADVANTAGE?

There is a second reason why the public AI environment is not a legitimate place to look for competitive business value. That reason is that public AI data is available to anyone.

If by some chance your corporation happened to find some important business data in the public AI environment, then ANY company can also find that very same data.

There is no business advantage gained when your competitors have the same data as you have. You have to assume that the competition is as smart as you.

Anything you can find in the public AI environment to gain business advantage, your competition can also find and use what they have found to their advantage.

There is then, no competitive advantage over your competition to using the public AI environment to your own advantage.

CONFIDENTIAL CORPORATE INFORMATION

There is yet another important reason why the public AI environment is the wrong place to look for business value. (And this is the most important reason of all.)

Consider the private data your corporation has that should never go beyond the walls of the corporation.

Your corporation has data about –

Proposals

Evaluations

Corporate strategy

Product acceptance

Budget

Sales results

Marketplace assessment

Human resources

Medical records

and so forth.

This data should NEVER find its way into the public domain. Having this data become exposed publicly can lead to lawsuits. Unnecessary lawsuits. And in some cases it may even be criminal to allow such data to be available publicly.

CONFIDENTIAL CORPORATE DATA AND BUSINESS VALUE

Ironically, the data that is the most confidential to the business is also the data that has the greatest chance of providing business value.

WHERE TO FIND BUSINESS VALUE IN AI

There is a solution to the conundrum that faces the corporation. If the public AI environment is not the place to find business value, where is the place to find it?

It needs to be recognized that there are two basic kinds of AI – public AI and private or corporate AI.

If company confidential data is placed on the corporate AI environment, the data never has to leave the confines of the corporation. In doing so the privacy of the data is protected. AI can be used within the walls of the corporation and there is no chance that the data will ever find its way into the public domain.

CONSEQUENCES

There are however some consequences of the need to do AI processing only on the corporate AI environment.

INTERNAL CORPORATE POLITICS

The first consequence is that doing massive amounts of processing on the corporate computer environment is not what the AI vendor has been telling management to do. The AI vendor operates under the assumption that the public domain is where the massive amounts of AI processing will be done. And the AI vendor has the ear of top management and the board of directors.

So there are internal corporate political obstacles to overcome.

LOCAL MANAGEMENT

A second challenge is that if AI processing is to be done within the boundaries of the corporation, AI processing must be managed and operated by local staff. The data administrators of the world are the best positioned employee to assume this role.

When AI was envisioned as a public AI centered world, large vendors sprang up and created LLM’s and infrastructure. There is little if any corporate support needed to use the public AI world.

But when it is recognized that the corporation must build and maintain its own AI environment, the task of building the LLM and the corporate AI infrastructure falls to the corporation.

And this is a nontrivial solution.

DIFFERENT SKILL SETS

One of the challenges is that the skill set of the classical data administrator is completely different from the skill set needed to manage the AI environment. The data administrator that has been with the corporation for 20 years is versed in the skills needed to run and manage the structured, transaction processing environment.

While those are valuable skills, when the proposition of the advent of the corporate AI arises, the corporation awakens to find out that an entirely new skill set is required.

Stated differently, computer professionals who have spent decades learning their craft wake up one day and find out that everything they have learned needs to be relearned in a completely different context.

This actuality is – naturally – disrupting. No one likes to hear this.

HANDS ON AND PUPPEERING

One of the fundamental reasons why the skill set of data administration needs to be relearned is that data management in the structured, transaction processing world is fundamentally different from the skill set needed for data management in the corporate AI world.

In the structured, transaction processing world, the data administrator operates in a very much hands on environment. If a data base needs to be monitored, the data administrator arranges that it be monitored. If a data base has an error, the data administrator fixes it. If a data base needs to have an index added, the data administrator adds the index.

In the structured, transaction processing world, the data administrator is very much a hands on individual.

But in the corporate AI world, the data administrator influences the system indirectly. In many ways the data administrator in the world of corporate AI is like a puppeteer. The data administrator controls the strings that shape the information, but the data administrator in the corporate AI environment has very few hands on tasks to do.

The data administrator in the corporate AI environment focusses his/her attention on the building and the management of the corporate LLM environment.

Now you know where to look to start to be able to find business value in your corporation.

In a word –

Business value is not to be found in the public AI environment

Business value is to be found in the corporate AI environment

The building and ongoing management of the corporate LLM is the key to finding and sustaining business value with AI

For more information about data architecture, data administration and the achievement of business value, please take a look at Bill Inmon’s Substack.

Bill Inmon’s Substack is the best source for contemporary thinking about data and data architecture. Authors and data architects from around the world have contributed their best thoughts.

And all monographs found on Bill’s Substack are free – the most loved word in the world.