AI has removed the hiding places at work

There is an uncomfortable workplace truth emerging around AI.

AI is not making people lazy. It may be exposing who was already waiting.

That is not meant as an attack. It is not a judgement on people who are learning, adapting or still building confidence with new tools. Every major workplace shift creates a learning curve.

But AI has changed something important. It has reduced many of the old reasons for not starting.

For years, knowledge work had plenty of acceptable hiding places:

• "I need to research this first."
• "I am waiting for a template."
• "I need someone to help structure it."
• "I will send a draft next week."
• "I just need to think about it properly."
• "I want to make sure it is polished before I share it."

Some of those reasons were valid. Some were not.

The difference now is that AI has removed many of the easy ones.

A first draft is available. A structure is available. A summary is available. A comparison is available. A meeting prep note is available. A sharper version of an email is available.

Not perfect. Not final. Not always correct. But started. And that changes the energy of work.

AI does not replace effort. It raises the floor. And once the floor rises, standing still becomes more visible.

The old hiding places were built into the workflow

Before AI, a lot of work could be delayed without looking like delay.

A person could say they were still researching. Still structuring. Still gathering inputs. Still waiting for a colleague. Still trying to find the right example. Still polishing the first draft.

In many cases, that was real work. Good work often does require thinking, research and reflection. But in practice, those phrases also created cover for hesitation.

The blank page was a blocker. The messy first version felt risky. The lack of structure slowed momentum. The fear of being wrong made people wait.

AI does not remove all of those challenges. But it changes how defensible the delay feels.

If a person can generate a rough outline in minutes, then "I did not know how to start" becomes less convincing. If a person can ask for questions to guide a customer conversation, then "I was not sure how to prepare" becomes weaker. If a person can summarise a long document and then verify the important points, then "I have not had time to read anything yet" becomes less acceptable.

The point is not that AI does the job. The point is that AI removes some of the friction that allowed the job not to begin.

What the evidence actually says

This is not a claim that AI magically improves every task. It does not. The research is more nuanced than that.

The Harvard Business School and Boston Consulting Group study on the "jagged technological frontier" found that consultants using GPT-4 completed 12.2% more tasks, completed them 25.1% faster and produced outputs rated more than 40% higher in quality on tasks inside AI's capability range.[1]

That is the optimistic part. AI can help knowledge workers move faster and produce stronger work when the task fits the tool.

But the same research also introduced a major warning. The impact of AI is uneven. Tasks that appear similar to humans may sit on different sides of AI's capability boundary. When users rely on AI outside that boundary, performance can suffer.[1]

That is the "jagged frontier". AI helps powerfully in some places. It misleads confidently in others. That matters because the real point is not "use AI for everything". The point is "use AI with judgement".

A separate study published in the Quarterly Journal of Economics examined more than 5,000 customer-support agents using generative AI assistance. It found a 15% average productivity increase, measured by issues resolved per hour, with substantial variation across workers.[2]

The largest gains were seen among less-experienced and lower-skilled workers. The study also found that the most experienced and highest-skilled workers saw smaller gains, and in some cases small declines in quality.[2]

That finding is important. AI can raise the floor. But raising the floor also changes expectations. If more people can get to "acceptable" faster, then standing still becomes more visible.

AI removes excuses, not accountability

The risk in this conversation is that people turn AI into a stick. "AI exists, so why is this not done?" That would be the wrong lesson.

AI does not remove complexity. AI does not remove workload. AI does not remove competing priorities. AI does not remove the need for human support.

But it does change what "blocked" means. Before AI, a person might be blocked because they had no structure, no draft, no starting point or no way to test an idea quickly. Now they often have a way to create a first version.

That does not mean the first version is good. It means the conversation can move. Instead of "I have not started", the expectation becomes "Here is a first version. Here is what I think is weak. Here is where I need input."

That is a better workplace pattern. It keeps the human collaboration. It keeps the need for feedback. It keeps the importance of context. But it shifts the individual responsibility from waiting to moving.

The new divide is movers versus waiters

The real divide at work is not simply between people who use AI and people who do not. That is too basic.

A person can use AI badly. They can generate more words without adding clarity. They can produce more slides without improving the decision. They can create confident summaries without verifying them. They can mistake activity for progress.

So the divide is not AI users versus non-users. It is between people who use AI to move and people who use AI to pretend they are moving.

One person uses AI to get unstuck:

• They create a first draft.
• They test a difficult argument.
• They prepare better for a meeting.
• They ask for counterpoints.
• They improve a customer note.
• They challenge their own assumptions.

Another person uses AI to create noise:

• More wording.
• More polish.
• More generic thinking.
• More confident filler.
• More activity that looks like progress.

The first person is using AI as leverage. The second person is using AI as theatre. That distinction will become more important.

The blank page is no longer the same obstacle

For many knowledge workers, the blank page used to be the hardest part. It created a particular kind of pressure. You had to start with your own structure. You had to decide what mattered. You had to make the first sentence work. You had to expose your thinking before it was ready.

AI changes that experience. The first version can now be created quickly. A weak version. A rough version. A version that needs challenge. But a version.

That matters because a weak first version is often better than no first version. It gives people something to react to. A proposal can be improved. A meeting agenda can be sharpened. A market summary can be corrected. A difficult email can be softened. A strategic argument can be pressure-tested.

AI does not remove the thinking. It gives the thinking something to push against.

What managers should start expecting

This should change what managers expect from teams. Not in an unreasonable way. Not "everything must be instant". But the baseline should move.

A rough draft in an hour may now be reasonable for many tasks. A meeting summary in minutes may now be reasonable. A first view on a market, customer or competitor may now be reasonable. A list of assumptions, risks and questions may now be reasonable before the meeting.

Not perfect. Started. That is the key word. The new expectation is not perfection. It is momentum.

• Bring the draft.
• Bring the thinking.
• Bring the assumption.
• Bring the question.
• Bring the comparison.
• Bring the first version.

Then let humans improve it. That is a powerful operating model. It also makes passive behaviour harder to hide.

The danger of faster theatre

There is a trap here. AI can make people look productive without being productive.

That is why managers should not simply reward output volume. More documents do not necessarily mean better work. More presentations do not necessarily mean better thinking. More meeting notes do not necessarily mean better decisions. More polished writing does not necessarily mean more value.

The real question is whether AI is improving movement:

• Are decisions moving faster?
• Are customers getting clearer answers?
• Are teams preparing better?
• Are junior people learning faster?
• Are managers spending less time fixing the basics and more time coaching judgement?
• Are teams surfacing risks earlier?
• Are people asking better questions?

If the answer is yes, AI is helping. If the answer is only "we now have more content", the organisation may simply be producing faster noise.

The human layer still matters

There is another risk in this conversation. We can become too excited about speed. Speed is useful. But judgement is still the differentiator.

AI can help draft the message. It cannot fully know the politics in the room. AI can summarise a market. It cannot replace commercial instinct. AI can sharpen a proposal. It cannot own the relationship. AI can help prepare for a difficult conversation. It cannot care on your behalf. AI can help create a first version. It cannot carry the accountability for the final decision.

That is why this is not a replacement story. It is an expectation story. The standard for starting has moved. The standard for judgement has not disappeared.

The leadership challenge

Microsoft and LinkedIn's 2024 Work Trend Index found that 75% of global knowledge workers were already using AI at work, and that 78% of AI users were bringing their own AI tools to work.[3]

That means AI adoption is already happening inside organisations, whether formal strategies are ready or not. This creates a leadership challenge.

If leaders ignore AI, adoption becomes private and inconsistent. Some people will use it well. Some will use it poorly. Some will use it invisibly. Some will overtrust it. Some will avoid it completely.

The better answer is to make the use of AI visible and disciplined. Ask people to show their process:

• What did you ask?
• What did it produce?
• What did you reject?
• What did you verify?
• What did you add?
• What changed because of your judgement?

That turns AI from a shortcut into a learning system. It also makes movement visible.

AI changes the excuse economy

Every workplace has an excuse economy. Some excuses are real. Some are habits. Some are inherited from old workflows. Some are simply fear dressed as process.

AI puts pressure on that economy. It does not remove the need for time. It does not remove the need for deep work. It does not remove the need for collaboration.

But it does remove some of the old hiding places. It is harder to claim there was no starting point. It is harder to claim there was no structure. It is harder to claim there was no way to prepare. It is harder to claim there was no way to test a first view.

That is a good thing if handled well. Because the goal is not to shame people into speed. The goal is to create a culture where people move earlier, ask better questions and improve work faster.

The question

So here is the question. Where has AI removed your excuse not to start?

Not where has it replaced you. Not where has it made you superhuman. Just where has it helped you move from "I will get to it" to "Here is the first version."

Because that may be the real workplace shift. Not that AI does the work. But that it makes waiting much harder to explain.

---

Sources

[1] Fabrizio Dell'Acqua et al., "Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of AI on Knowledge Worker Productivity and Quality", Harvard Business School working paper with Boston Consulting Group, 2023. [2] Erik Brynjolfsson, Danielle Li and Lindsey Raymond, "Generative AI at Work", Quarterly Journal of Economics, 2025. [3] Microsoft and LinkedIn, "2024 Work Trend Index: AI at Work Is Here. Now Comes the Hard Part", 2024. [4] Shakked Noy and Whitney Zhang, "Experimental evidence on the productivity effects of generative artificial intelligence", Science, 2023.