Why the widespread adoption of AI by consumers says nothing about its future in the workplace

For two years now, we have been hearing that generative AI is everywhere, that it will transform the way we work, and that its widespread adoption by the general public will inevitably lead to its rapid deployment in businesses. However, in reality, its adoption in the professional world remains tentative. This discrepancy raises questions: if everyone is using it in their personal lives, why aren’t organizations embracing this movement with the same enthusiasm?

The reality behind the talk of the “massification” of uses and the consumerization of enterprise software is much more nuanced. The recent history of technology reminds us that there is a gulf between spontaneous individual adoption and structured collective integration, and that this gulf cannot be bridged by a simple fad.

In short:

• Personal adoption of generative AI does not automatically translate into professional use, as the context, constraints, and objectives in business differ from those in the private sphere.
• The history of previous technologies (enterprise social networks, remote work, Web 2.0) shows that tools that are popular in the personal sphere may fail to deliver value in a professional context.
• Consumer uses of AI are often ad hoc, simple, or convenience-driven, which does not meet the investment criteria of businesses seeking measurable and sustainable gains.
• Barriers to business adoption include security, compliance, interoperability, reliability of results, and the heterogeneity of benefits depending on user skill levels.
• Indicators such as low paid subscription rates and a decline in usage after the novelty effect suggest that generative AI is still perceived as an accessory tool rather than an essential performance lever.

The temptation to transpose the consumer market to the workplace

It’s a classic mistake, already seen in other waves of digital transformation: believing that what people do in their personal lives, they will do in the same way at work. For example, in the mid-2010s, we were already hearing executives complaining that their employees weren’t “digital enough” while their customers were. I pointed out the absurdity of this bottom line: in many cases, the “non-digital” employee and the “digital customer” were one and the same person, but stepping through the office door changed everything (Are your employees really hopeless at digital ?).

The simple act of walking through the office door changes the context: the tools, procedures, compliance requirements, and above all, expectations are no longer the same. Behavior that seems natural in a private setting can become inappropriate or impossible in a professional setting. The environment plays a much more structuring role than the supposed individual digital culture.

The illusion of social media in business

We saw the same error in judgment with “Web 2.0” and its internal equivalent, namely corporate social networks, at the turn of the 2008–2010 period. The argument was crystal clear on paper: young people spend their lives on Facebook, so they will excel at animating internal communities, sharing experiences, and collaborating. The reality was very different, and it’s fair to say that the results fell far short of the promises, even if maturity and culture were only one of many factors contributing to the failure (The rise and fall of enterprise social networks and Social collaboration isn’t lacking tools, it’s lacking permission).

In the workplace, it is often the more experienced employees, with knowledge to pass on and experience of working in a team, who have taken advantage of these tools. Younger employees, on the other hand, were very comfortable with informal or playful exchanges, but found themselves at a loss when faced with professional interactions that required rigor, clarity, and relevance. The tool was the same, but the context completely transformed its use.

When domestic use collides with professional reality

The remote work imposed during COVID provided another demonstration. Yes, almost everyone had a computer and internet at home, but that did not mean that everyone was ready for intensive and structured use. Many people worked on a laptop placed on a corner of a table or on their lap, with personal tools that were ill-suited to the security and productivity requirements of a business and, let’s face it, for such basic uses that at home, the phone or tablet meant that the computer was usually left in a drawer (Digital has entered people’s homes…but not in the way you might think).

When the time came to install and use a VPN, connect to a business tool, or make specific settings, the illusion of “widespread digitization” was shattered. Familiarity with simple digital uses does not prepare you for the complexity of professional environments.

Playing with technology is not using it seriously

Today, we have the same confusion with generative AI. Much of its use by the general public is for convenience or entertainment: translating a text, rephrasing a message, preparing a recipe or a trip, or getting a quick summary. These are real time savings, but they remain marginal in terms of added value, especially when evaluated on the scale of a complete business process.

However, for a business to invest, the benefits must be clear, measurable, and sustainable. A tool that only provides a one-off or anecdotal improvement will not be a priority, even if its users appreciate it on a daily basis. This difference in logic also partly explains why the enthusiasm of the general public does not automatically translate into professional adoption.

What the data says about usage

A report by the Harvard Business Review confirms this disconnect (How People Are Really Using Gen AI in 2025). Of the 100 main use cases identified, 31% relate to “personal and professional support”, and the top 10 are dominated by highly personal goals: organizing one’s life, finding purpose, improving health, and receiving emotional support. These are all perfectly legitimate areas in the private sphere, but they do not correspond to the investment priorities of senior management, despite their rhetoric on quality of life at work.

This does not mean that these uses have no value in business, but their direct impact on key indicators (productivity, revenue, customer satisfaction) is often low. What, in an individual’s daily life, is an improvement in comfort does not automatically translate into value creation for the organization.

The wallet test

Another indicator can be used to measure the actual maturity of a technology’s use: willingness to pay. ChatGPT currently has around 700 million active users each week, but fewer than 10 million subscribe to a paid service, representing less than 1.5% (Number of ChatGPT Users (July 2025)).

If generative AI had become essential to everyday life, we would see this rate climb much higher. The fact that the vast majority are satisfied with the free versions shows that, for many, the tool remains useful but not vital. Recent restrictions on free offers and price increases will serve as a test (The Enshittification of Generative AI): if the conversion to paid services remains marginal, this will confirm that AI is more of a convenience than a critical tool and that its adoption by the general public does not necessarily predict its future success in business.

The hidden side of personal adoption

The gap between private use and professional adoption is not only due to the nature of the needs, but also to the mechanics of integration. An individual can adopt a tool in a matter of seconds, without worrying about security, privacy, or interoperability. A business, on the other hand, has to deal with constraints on another level, such as controlling data storage, ensuring GDPR compliance, and integrating the new tool into its information system without creating new vulnerabilities (Why enterprise AI can’t keep up with consumer AI: beyond ChatGPT, a more complex reality).

Added to this is an invisible cost: that of verification and reliability. In the workplace, a gross time saving is not a net gain if each output has to be proofread, corrected or completed. The sum of the number of augmented employees does not necessarily create a more efficient organization (AI in the workplace: going beyond augmentation to actually transform).

The skill paradox

Another limitation that is rarely discussed is that AI delivers better results to those who already have a good grasp of the subject matter. An expert will know how to ask the right questions, interpret the answers correctly, and use them effectively, while a novice may obtain results that appear impressive but are inaccurate or unusable.

This uneven leverage poses a problem at the business level: the same tool can be a tremendous accelerator for some and an ineffective gadget for others. However, to justify a massive investment, the impact must be consistent and measurable across the organization.

The weak signals of disinterest

The low subscription rate is not the only factor to consider. There is also a decline in frequency of use once the novelty effect has worn off (Novelty effect and The Evolving Usage of GenAI by Computing Students), and the fact that much of the use remains outside formal processes and is therefore invisible in performance measurements (Enterprise AI trends in 2025: what’s real vs. pure hype and The hidden economics of AI: balancing innovation with reality).

Furthermore, a significant proportion of interactions with AI are used to solve one-off problems rather than to transform a workflow (Optimized by AI, Undermined by Design: What’s Breaking Work Behind the Interface). As long as AI remains confined to this role of “occasional troubleshooter” it cannot claim the status of a structuring tool.

Bottom line

Recent history has shown that mass exposure of a population to a technology does not guarantee its maturity or its ability to be adopted in a professional setting, and even when it does, businesses are rarely ready to embrace these new uses in a way that will have an impact.

There is a gap between playing with a tool and turning it into a strategic lever, and this gap often takes a long time to bridge, if it is bridged at all. As long as generative AI remains confined to simple, ad hoc uses that are mostly basic enough to be satisfied with free offerings, it and its users will not cross this threshold. 

And as long as the cost of integrating and deploying it in businesses remains high, executives will probably be right not to rush, or rather, to hurry slowly.

Image credit: Image generated by artificial intelligence via ChatGPT (OpenAI)

Related articles:

Why use cases don’t work with freeform technologies Why enterprise AI can’t keep up with consumer AI: beyond ChatGPT, a more complex reality Generative AI in the workplace: revolution or illusion? AI heading for an economic dead end? Generative AI: your organization is worth more than your tools The limits of technology-driven transformation