Average work has become harder to defend

For years, effort was easy to confuse with value.

A document took three days. A deck took a week. A proposal went through several versions. A market summary took hours. A meeting note had to be written from scratch.

Because it took time, it felt valuable.

That logic is starting to break. AI has compressed the time it takes to produce basic knowledge-work output.

Not perfect output. Not strategic output. Not deeply contextual output. But acceptable output. A summary. A first draft. A comparison. A structure. A meeting note. A polished email. A presentation outline.

These things used to carry more perceived value because they required friction. Now the friction is lower.

So the question changes. It is no longer only "How long did this take?" It is "What did the human add?"

That is why average work has become harder to defend.

AI has raised the baseline

AI has not made great work less valuable. It has made average work easier to spot.

This distinction matters. The strongest work still requires judgement, context, originality, customer understanding, commercial instinct and accountability. AI can help with all of those things, but it cannot fully own them.

What AI has changed is the baseline. A basic first draft is now easier to produce. A passable summary is easier to create. A neat email is easier to polish. A generic proposal is easier to assemble. A presentation structure is easier to build.

That does not mean the work is excellent. It means the minimum standard has moved. When everyone has access to tools that can generate acceptable work quickly, acceptable stops being impressive.

The premium moves higher. It moves to the thinking behind the work.

The new question is no longer "How long did this take?" It is "What did the human add?"

The evidence supports the baseline shift

The productivity evidence around generative AI is strong enough to take seriously, but nuanced enough to avoid hype.

In an experiment published in Science, Shakked Noy and Whitney Zhang found that ChatGPT reduced average completion time by 40% and increased output quality by 18% on professional writing tasks.[1]

That matters because writing is a core layer of knowledge work. Most modern work is not only done through spreadsheets, systems or meetings. It is done through language: emails, briefs, summaries, proposals, reports, policies, presentations, meeting notes, customer updates.

If AI can help people write faster and improve average quality, it changes the value of basic written output.

A separate study published in the Quarterly Journal of Economics looked at more than 5,000 customer-support agents using generative AI assistance. The study found a 15% average productivity increase, measured by issues resolved per hour, with the largest gains among less-experienced and lower-skilled workers.[2]

That is important. It suggests AI can raise the floor. People with less experience can perform closer to stronger performers in some structured tasks.

But that also changes what counts as differentiated. If AI helps more people reach "good enough" faster, then "good enough" can no longer be the ambition.

The jagged frontier still matters

There is a danger in taking the productivity evidence too far. AI does not improve every task.

The Harvard Business School and Boston Consulting Group study on the "jagged technological frontier" found that consultants using GPT-4 performed significantly better on tasks inside AI's capability range. But it also found that performance could worsen when people used AI outside that range and trusted it too much.[3]

That point is essential. AI can raise the baseline on many tasks. But it can also make weak work look stronger. It can produce confident answers that are incomplete. It can create polished writing that lacks insight. It can generate summaries that miss critical context. It can offer recommendations that sound plausible but do not fit the real world.

So the conclusion is not "AI makes work better." The better conclusion is "AI makes basic output easier to produce, which increases the importance of human judgement." That is the shift.

Average work used to hide behind friction

Average work was often protected by effort. Not because people were dishonest. Because the process was genuinely slower.

A person had to gather information. They had to structure it. They had to write it. They had to edit it. They had to format it. They had to make it look presentable. That took time.

And because it took time, the work often carried a sense of value before anyone had properly tested its usefulness. A long document felt substantial. A polished deck felt considered. A detailed summary felt diligent. A carefully worded proposal felt strategic.

AI challenges that. It can help create length. It can help create polish. It can help create structure. It can help create confidence. But those things are no longer enough.

A long document can still say very little. A polished deck can still lack insight. A detailed summary can still avoid the real issue. A carefully worded proposal can still be generic.

Average work is becoming harder to defend because the old signals of effort are losing power.

The new premium is contribution

If basic output is easier to produce, the premium moves to contribution:

• What did the person notice?
• What did they challenge?
• What did they understand about the customer?
• What did they know from context?
• What did they add from experience?
• What risk did they identify?
• What decision did they make easier?
• What trade-off did they clarify?

That is the human value. Not simply producing the document. Moving the thinking forward.

This is especially important for leaders. In an AI-enabled workplace, managers should be careful not to reward the wrong thing. More content is not always more value. More slides do not always mean clearer decisions. More summaries do not always mean better understanding. More analysis does not always mean stronger judgement.

The question should move from production to contribution.

The problem with "good enough"

AI is very good at helping people reach "good enough". That is empowering. It helps people start faster. It reduces blank-page anxiety. It improves structure. It creates momentum. It helps teams produce first versions faster.

But "good enough" is also dangerous. Because once everything is more polished, polish becomes less impressive.

A well-written document is not automatically a useful document. A neat presentation is not automatically a strong recommendation. A confident summary is not automatically an accurate summary. A fast answer is not automatically the right answer.

The risk is that average work now arrives looking better than before. Cleaner. Sharper. More structured. More confident. But still average.

That makes the manager's job harder. It also makes judgement more important.

The real risk is polished mediocrity

The old version of average work was easier to identify. It looked rough. The thinking was messy. The structure was weak. The writing was uneven. The gaps were visible.

That made it easier to coach. A manager could see where the person was struggling. A senior colleague could redline the work and explain what was missing.

AI changes that. The new version of average work can look professional. It can have a clean structure. It can have strong grammar. It can have confident phrasing. It can have a neat executive summary. It can have sensible-looking bullets.

But it may still lack substance. The assumptions may be weak. The argument may be generic. The customer context may be missing. The data may be unverified. The recommendation may be obvious. The trade-offs may be underdeveloped.

This is polished mediocrity. It is one of the biggest risks of shallow AI adoption. Not bad work. Average work that looks too good to challenge quickly.

What leaders should ask instead

Leaders should change the questions they ask when reviewing work. The question should not only be "Is this well written?" That bar is too low now.

Better questions are:

• What did we learn that was not obvious?
• What assumption was challenged?
• What customer context was applied?
• What decision does this help us make?
• What trade-off is being recommended?
• What risk has been surfaced?
• What evidence supports this view?
• What would a sceptic say?
• What did the human add that AI would not have known?

That last question matters most. Because in an AI-enabled workplace, the value of human work will be judged less by production and more by contribution. The person who simply produces something will become less differentiated. The person who improves the decision will become more valuable.

What this means for individuals

This shift is not only a leadership issue. It matters for every professional. If AI can help produce the basics, then your value needs to sit above the basics.

That does not mean everyone must become a visionary strategist. It means people need to be more deliberate about the human layer they add:

• Bring context.
• Bring a point of view.
• Bring customer reality.
• Bring evidence.
• Bring better questions.
• Bring trade-offs.
• Bring judgement.
• Bring courage.

Do not only bring the output.

The more AI becomes embedded in the workplace, the more dangerous it becomes to hide behind activity. "I produced the deck" is not enough — did the deck clarify the decision? "I wrote the summary" is not enough — did the summary surface what mattered? "I drafted the proposal" is not enough — did the proposal reflect the customer's real problem?

That is the new standard.

Average work is not a people problem

This is not about attacking people. Average work is often not the result of laziness. People are busy. People are stretched. People are interrupted. People have limited time. People have to operate with incomplete information.

In many organisations, average work is a system outcome: too many meetings, too many priorities, too little thinking time, too much urgency, too little clarity on what good looks like.

AI does not automatically fix those problems. But it does change the starting point. It gives people more leverage. That is empowering. It also means the target has to move.

The goal cannot be "I produced something." The goal has to be "I moved the thinking forward."

The organisation has to raise its standard carefully

There is a responsible and irresponsible way to respond to this shift.

The irresponsible way is to demand more from people simply because AI exists. More output. More speed. More documents. More expectations. That will burn people out and create noise.

The responsible way is to raise the quality bar. Use AI to reduce the friction of starting. Use AI to reduce repetitive drafting. Use AI to create faster first versions. Use AI to support preparation.

Then use human time for the work that matters more: judgement, coaching, decision-making, customer conversations, strategy, learning, relationship-building.

AI should not be used to flood organisations with more average work. It should be used to create space for better work.

The real shift

The real shift is not that AI replaces people. It is that AI changes what we should value from people.

Less time spent proving you worked hard. More focus on whether the work helped:

• Did it clarify the decision?
• Did it improve the customer conversation?
• Did it identify the risk?
• Did it simplify the complexity?
• Did it generate confidence?
• Did it move the business forward?

That is the work that becomes more valuable. Not because AI cannot help with it. But because AI alone cannot own it.

The question

So here is the question. What kind of work used to feel valuable because it took time, but now needs to prove real value? The first draft? The market summary? The meeting note? The proposal? The polished deck? The quick analysis?

Where are you seeing average work become harder to defend?

Because AI is not making great work less important. It is making the difference between average and valuable much harder to ignore.

---

Sources

[1] Shakked Noy and Whitney Zhang, "Experimental evidence on the productivity effects of generative artificial intelligence", Science, 2023. [2] Erik Brynjolfsson, Danielle Li and Lindsey Raymond, "Generative AI at Work", Quarterly Journal of Economics, 2025. [3] 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. [4] Microsoft and LinkedIn, "2024 Work Trend Index: AI at Work Is Here. Now Comes the Hard Part", 2024.