The AI Chasm - Adoption vs. Acceptance

When assessing the success and value of a software product, one metric frequently measured is adoption. In other words, how many people are using or engaging with the product/technology?

 

For example, one measure of adoption is the number of people using it. Sometimes, this gets reported by age group, country, etc.

 

It’s easy to misinterpret adoption with acceptance. People may be using something, but only because they are forced to, as in the case of enterprise applications.

 

Acceptance can be seen in various ways. In some cases, it is a formal acceptance or rejection. In other cases, acceptance is a foregone conclusion, such as when a contract has already been signed or a purchase has already been made.

 

Another case of assumed acceptance is seen in SaaS applications, where one day you wake up to a new version of the application. Sometimes there is a notice, sometimes not. It all depends on the vendor.

 

Acceptance testing may be used for formal acceptance or rejection, or it may be used to validate how good or bad something is, perhaps to find the gaps between how the application works and how the business works.

 

Adoption and acceptance are related, yet different. Just because you use something, even to a great degree, doesn’t mean you fully accept it or like it. Forced adoption is even worse because when forced to use a product, people can comply yet not use it well.

 

One interesting way to view adoption was presented many years ago in a book by Geoffrey Moore called “Crossing the Chasm.” The book was about how to market disruptive technologies. Well, AI is certainly disruptive!

 

The main takeaway from the book is that you have people who are innovators and early adopters. After that comes the early majority. Between the early adopters and the early majority is a big gap (the “chasm”) in which the messaging must change to win the early majority over. I think we have already crossed that chasm. Next comes the late majority. I think this is where we are crossing into now.

 

Customers vs Users

 

When discussing adoption or acceptance, it is important to understand the difference between a customer and a user. In some cases, the two roles may be played by the same person.

 

A customer is the person who makes the buying decision. A user is the person who actually uses the product or service. In enterprises, it’s very common for senior management to make the buying decision, then force the rest of the organization to adopt and use the product or service.

With AI, an individual might subscribe to ChatGPT, at which point they are both customer and user. A company may make a multi-million-dollar investment in AI, which makes them the customer and everyone else in the organization a user, whether they like it or not.

 

AI is Getting Mixed Reviews

 

Judging by all the buzz, one might think that AI is at a very high rate of adoption. However, the most recent data (January 2026) show that only 13.3% of people worldwide are using AI with any regularity. [i]We can assume that “using” AI can be anything from a chatbot to agents. My suspicion is that a significant part of this 13% is due to chatbot use and the creation of funny videos.

 

Interestingly, 56% of people in the USA use AI, with 28% reporting they use AI weekly. 31% of Americans have never used AI.

 

Daily use is reported by 12%, use multiple times daily is only 6% in the USA.

 

With the speed of adoption, treat these data as a growing baseline. The actual numbers are most certainly increasing. The data also takes time to gather, so the actual numbers are from mid to late 2025.

 

However, even with that caveat, the numbers seem much lower than the tech leaders and billionaires would indicate. Some have speculated that the picture is inflated to help drive investments.

 

On the heavy side, 88% of companies report using AI, 79% of companies are using generative AI, and a whopping 96% of development teams say they are using AI, The stunning part is that 24.6% of those surveyed say they are using software composition analysis to verify the security of code suggestions from AI tools!

 

We have other data about developers’ use of AI.

 

In the StackOverflow 2025 Developer Survey, 80% of developers reported using AI. However, trust in the accuracy of AI has fallen from 40% in previous years to just 29%. Favorability in AI decreased from 72% to 60%.

 

66% of developers say they are spending more time fixing "almost-right" AI-generated code. 75% said they would still ask another person for help when they don’t trust AI’s answers.[ii]

 

A Quiet Rebellion is Happening

 

This is where things are starting to get interesting.

 

There appears to be a growing amount of push-back (a.k.a. “rejection” or “non-acceptance”) of AI by people, both individually and corporately.

 

According to an article on Cybernews.com dated March 17, 2026:

 

“Just 26% of voters said they feel positively about AI, compared with 46% who hold negative views. In fact, the only topics with a lower net positive rating than AI in the NBC News survey were the Democratic Party and Iran. Even the US Immigration and Customs Enforcement, the militarized agency enforcing its brutal deportation program and shooting unarmed civilians, is viewed more positively, the poll found.”[iii]

 

Part of the rejection of AI is being seen as people who use AI (such as at work), but who reject being forced to use it. These people feel that it is diminishing their experience and judgment as they do their work. Indeed, in a Harvard study, the people who exhibit the most value with AI are those who don’t rely on it for everything but treat it as an assistant that requires close monitoring.

 

A research team at Harvard studying workers at Procter & Gamble and Boston Consulting Group looked at where AI is most effective in increasing productivity and performance. It also examined where humans still have the advantage.

 

One interesting finding from their research is that while individuals using AI might gain speed and performance, AI-enabled teams see the greatest gains. Just replacing humans with AI might not be the magic fix that some companies are betting on.

 

At first, this may seem like a contradiction in AI messaging, which is meant to let AI do all the analytical work. But when one considers that AI can be biased, make mistakes, and even lie, it makes sense that people still need to evaluate what it tells them.[iv]

 

The feeling that AI is eroding skills hits at the very heart of work's value. Many people work because they find satisfaction in it. When that satisfaction is taken away because technology can “do it better,” people start to reject it.

 

There is also the great desire for that which is authentic. Whether it be art, music, writing, or interacting with an AI-powered customer service agent, people want to connect with the soul of another human being. People are asking more and more, “Is this real or AI?” That is a question we should all be asking.

 

This is another paradox. On the one hand, people are told to use AI. “Don’t fall behind,” they are told. On the other hand, a growing number of people don’t like AI. Even LinkedIn has reportedly started detecting and penalizing AI-generated articles!

 

YouTube has also recently started demonetizing channels that promote heavy AI-generated content. The interesting twist here is that YouTube is using AI to detect AI content, and sometimes it gets it wrong. An example is animated video content. It’s hard to tell AI from human-created content at times.

 

Why are these platforms taking these actions? Personally, when I see obvious AI-generated content, it makes me feel like the person didn’t want to put the work in to craft an authentic piece. Or, perhaps they are unable to think for themselves, so they need AI to think for them.

 

Sure, there is a time and place for AI to show value, such as in research and idea generation. But care must be taken, because AI will omit things, misattribute things, and lie about others.

 

Attorneys are being sanctioned for citing non-existent cases, based on AI input.

 

A Few Other Data Points

 

Several recent college commencement speakers have been booed when they mentioned AI. Yes, it got ugly. The speakers were shocked at the reaction. Apparently, they did not realize their audience is fighting against AI for entry-level jobs.

 

AI data centers are being built across the USA, but people in those areas do not want them in their backyards – or anywhere near their backyards. They consume huge amounts of water, electricity, and other resources while generating heat and noise. Farmland is being destroyed. The tech companies have been given special tax exemptions, which permit them to take without giving back to the communities. Instead, residents are seeing their utility bills increase, not by a little, but by multiples.

 

Companies are radically shedding talented people to invest in AI and automation. The big question is: who will buy things if the working middle class, and even the upper-middle class, gets hollowed out? Elon Musk recently said that people should not worry about savings. What does he know that we don’t?

 

Google is upending the world of Internet searches by making the AI response the first one seen. That impacts both the searchers and content creators. Some searchers would like to actually browse through the search results. Yes, that is still possible, but AI is changing Search Engine Optimization so radically that the results you really want to see get buried.

 

For the content creators hoping that Google will rank their web content highly, this is bad news. A large majority of that search traffic has disappeared, with AI results taking its place. However, people do not fully trust the AI results, which is understandable given what we know about AI and accuracy. According to a Pew Research report from October of 2025, “Americans who have seen AI summaries in search results are lukewarm about their value. One-in-five say they find the information extremely or very useful, 52% say it’s somewhat useful, and 28% say it’s not too or not at all useful.” Only 6% report finding the AI results extremely helpful.[v]

 

I decided to ask AI about this. Here is what it told me. “Many people express wariness about AI's potential impact on search engines, as they are concerned about how it may change their information-seeking habits and the reliability of the results. While AI is reshaping search behaviors, traditional search methods still hold a strong preference among users.” Now, the question is, “Do I trust that?”

 

Where is All This Heading?

 

We are very rapidly heading for a world where both acceptance and adoption are going to be out of our hands as individuals, and even corporately.

 

Another pertinent book from the past (2004) is “The Inmates are Running the Asylum” by Alan Cooper. In that book, Cooper argues that the business executives who decide to develop complex technologies and products are not the ones in control of the technology used to create them. The book's context was PC-based software, but it also extends to other software-controlled devices. Following the book’s title metaphor, the inmates are the developers who create features but don’t really understand usability factors. The result is that usability often gets lost in the shuffle.

 

In the “AI arms race,” as some have called it, the tech giants would be the inmates. With AI, the concern is less about usability and more about safety and purpose.

 

I use AI for a variety of purposes, but I stop short of letting it think for me. I also refuse to shape my content around what I think AI wants to see, as with web search. I know I’m sacrificing speed for authenticity.

 

For example, this article has taken me at least three days to write. I probably thought about the concept for several days before I started typing. But what you are reading now is not “AI slop.” These are my thoughts and words, based on my experience and perspective.

 

For software testers and others, my encouragement is that while AI is a great tool, there are still many things it can’t do. Even at Artificial General Intelligence (AGI), AI might not be able to match human reasoning and experience. Assuming, that is, that we do not over-rely on AI to the point we forget our fundamental knowledge and skills!

 

I hope this article prompts some thought around how we, as knowledge workers, can use AI and benefit from it without losing our souls, our motivation, and our life’s work!

 

If you need assistance in defining a team-based AI QA and test strategy for your organization, contact me at the menu link above.