Asher's Blog

The most dramatic story about artificial intelligence is that it will take your job. I think that story is so dramatic that it can distract us from the more immediate danger.

AI does not have to replace you to reduce your value. You may keep your job title. You may continue to sit at the same desk, attend the same meetings and receive the same salary. Yet your relative value may already be falling because somebody else, doing similar work, has learned to use intelligent tools to produce better results in less time. The company does not need to fire you for this change to matter. It may simply give the more important assignments to somebody else. It may promote that person more quickly. It may decide that one AI-enabled employee can now do work that previously required two or three people. It may continue to employ you, but stop seeing you as essential to its future.

Replacement is an event. Reduction in value is a process, and the process has already started.

This is not a reason to panic. In fact, the beautiful thing about this period is that almost anybody can learn to use AI. You do not need to have studied computer science. You do not need to become a machine-learning engineer. You do not even need to be the youngest person in the room. You need curiosity, humility, good judgment and the willingness to practise until the tools become part of how you work.

The opportunity is not merely to protect the person you already are. It is to recreate yourself.

Your benchmark has changed

For a long time, employees compared themselves with other employees who had similar education, experience and access to resources. Today, that comparison is incomplete. A lawyer using AI carefully can review and organise information faster than a lawyer who refuses to use it. A salesperson can research an account, prepare questions, summarise a call and draft a follow-up in a fraction of the previous time. A marketer can study customer feedback, generate several campaign directions and test messages without waiting for a large team. An engineer can use AI to understand an unfamiliar codebase, draft tests and investigate errors.

The advantage does not come from asking a chatbot one casual question. It comes from redesigning the work.

This is why your output can remain exactly where it was and still become less valuable. Value is relative to the best credible alternative available to the customer or employer. If a report used to require two days and now somebody can produce a stronger, properly verified version in three hours, the market value of the old two-day process has changed. If a team needed five people to deliver a service and a smaller team can now deliver it faster, the economics of that work have changed. Your effort may be sincere, but the benchmark does not reward effort alone.

Research is beginning to show the size of this difference. In Generative AI at Work, Erik Brynjolfsson, Danielle Li and Lindsey Raymond studied more than 5,000 customer-support agents and found that access to a generative-AI assistant increased productivity by about 14 percent on average, with substantially larger benefits for less experienced workers. The technology did not remove every agent. It changed what an agent could produce within an hour. That alone changes how companies assess staffing, training and performance. (NBER)

In another experiment involving consultants, researchers from Harvard Business School, MIT, Wharton and Boston Consulting Group found that people using GPT-4 completed more tasks, worked about 25 percent faster and produced markedly higher-quality work on tasks that fell within the technology’s capabilities. But the same study also found that AI users performed worse on a task outside that capability frontier. The researchers called it a “jagged technological frontier”: AI can be remarkably strong in one area and unreliable in another that appears similar. (Harvard Business School)

That second finding is as important as the first. The future does not belong to the person who blindly copies whatever AI produces. It belongs to the person who knows when to use it, how to question it, how to verify it and when to set it aside.

You are competing with people who use AI well

People sometimes frame the issue as human beings versus machines. That is not the most useful way to see it. In the near term, most people will not lose an opportunity directly to a machine. They will lose it to another person or company that uses machines more effectively.

A small Nigerian business that uses AI to answer enquiries quickly, analyse sales patterns, prepare proposals and follow up consistently can take customers from a larger but slower competitor. An African professional who learns to combine local knowledge with global intelligent tools can compete for work that once appeared inaccessible. A young company can now create research, software, design and customer-support capacity that would previously have required far more capital.

This matters especially in Africa. We already operate with constraints: expensive capital, infrastructure gaps, small teams and customers with limited purchasing power. We cannot afford to preserve inefficient work simply because it is familiar. AI can help us overcome some disadvantages, but only if we use it to increase the standard of our output. If the rest of the world uses AI to become faster while we use it only to generate generic social-media captions, the productivity gap may widen rather than close.

I have had to push some people on my team to use these tools more seriously. Some are millennials; some are Gen Z. It surprised me because much of this information is public. The tools are available, tutorials are everywhere and people are demonstrating useful workflows every day. I should not always have to be the person telling a younger colleague what is possible.

But this has also taught me not to confuse digital familiarity with AI fluency. Being young does not automatically make somebody an effective user of new technology. Knowing how to use social media is not the same as knowing how to interrogate a model, provide useful context, evaluate an answer and turn it into a dependable process. Age is not the deciding factor. Curiosity is. Initiative is; the habit of learning without being chased is.

An employee who waits to be formally trained on every public tool is already communicating something about their value. They are saying that somebody else must first discover the opportunity, interpret it and bring it to them. A valuable employee should increasingly be able to say, “I found a better way to do this. I tested it safely. Here is the improvement in time, quality or cost.” That is ownership.

Access is not an advantage; application is

Almost everybody with an internet connection can access some form of AI. That means access itself will not remain a durable advantage. The advantage is in what you can make the system do consistently.

Ethan Mollick, in Co-Intelligence, encourages people to invite AI to the table in the work they do, while remaining the human responsible for the outcome. I like that balance. If you never bring AI into the process, you will not discover where it can help you. If you surrender the whole process to it, you may produce fast nonsense and damage the trust people have placed in you.

Using AI effectively requires more than prompting; it requires at least five levels of development.

The first level is use. You stop discussing AI only as an abstract trend and begin using approved tools on real work.

The second level is verification; you check facts, calculations, citations, assumptions and tone. You understand that fluency is not the same as truth.

The third level is integration. Instead of opening a chatbot occasionally, you connect AI to a repeatable part of your workflow: research, analysis, documentation, customer support, coding, planning or quality assurance.

The fourth level is redesign. You stop asking how AI can make the old process slightly faster and ask whether the process should exist in its current form at all. Perhaps a report should be generated continuously instead of manually at month-end. Perhaps customers should receive answers immediately rather than waiting for an employee to assemble information from five systems.

The fifth level is direction. You build or coordinate intelligent systems that can execute substantial work within clear boundaries, while human beings retain judgment, accountability and control.

Many people will remain at the first level and believe they are prepared. They will use AI to polish an email and say they are now AI-enabled. Meanwhile, somebody else will redesign an entire workflow, reduce errors, shorten turnaround time and produce measurable commercial value. Both people have access to the same technology, but they do not possess the same advantage.

Do not automate mediocrity

There is another danger. AI can help an excellent worker produce more excellent work, but it can also help a careless worker produce more careless work at scale.

If you cannot define the problem, AI may help you answer the wrong question faster. If you do not understand your field, you may not recognise a confident fabrication. If your taste is weak, you may accept the first generic output. If you lack integrity, the technology can help you disguise work you do not understand. And if you paste confidential company or customer information into an unapproved system, your attempt to appear productive may expose the organisation to serious risk.

The consultant experiment I mentioned earlier makes this plain: AI improved performance within its capability frontier, but dependence on it became a disadvantage outside that frontier. Judgment has therefore become more valuable, not less. Domain expertise has become more valuable. The ability to frame a problem, notice an inconsistency and take responsibility for the final decision has become more valuable.

The wrong response to AI is refusal. Another wrong response is surrender; you must remain the accountable person. If you send an incorrect analysis to a customer, “AI wrote it” is not a defence. If an automated system discriminates, leaks data or makes an unlawful decision, the organisation remains responsible. A serious professional uses AI as leverage, not as an excuse.

A practical way to reinvent your work

If you are wondering where to start, do not begin by trying every new tool. Start with your actual responsibilities.

Write down the work you performed over the last two weeks. Separate it into four categories: repetitive work, analytical work, relational work and judgment-intensive work. Then identify one recurring task that consumes meaningful time but has a clear output. It could be preparing a weekly report, analysing support tickets, drafting proposals, documenting meetings, checking reconciliations or conducting customer research.

Measure the current process. How long does it take? How many people touch it? What mistakes commonly occur? What does a good result look like? Without a baseline, you cannot honestly claim that AI made you more productive.

Next, use an approved tool to improve one part of the workflow. Give it the right context, examples, constraints and evaluation criteria. Do not paste sensitive data into a public model without permission. Review the result yourself or with somebody who understands the domain. Document what worked. Then repeat the task until the improvement is reliable rather than accidental.

After that, raise the standard. The time saved should not merely become more idle time. Use it to speak with customers, solve a harder problem, investigate an opportunity, improve quality or support the team. Productivity is not simply producing the same mediocre result more quickly. It is converting saved time into greater value.

Finally, teach someone else. When you can explain the workflow, its limitations and the safeguards around it, you understand it more deeply. The organisation also becomes less dependent on private tricks that disappear when one employee leaves.

For leaders, there is a corresponding responsibility. You cannot demand AI adoption while providing no approved tools, no security policy, no training and no clarity about acceptable use. Leaders must create a safe environment for experimentation, protect customer data, establish human review for consequential decisions and measure outcomes rather than theatre. Buying licences does not transform a company. Work changes when habits, processes, incentives and expectations change.

What will remain valuable

As AI makes some outputs cheaper, the human qualities surrounding those outputs become more important.

Problem selection matters because a perfect answer to an irrelevant question creates no value. Judgment matters because the system does not carry the consequences of a decision. Trust matters because customers need to know who will be accountable when something goes wrong. Taste matters because generating ten options is easy; knowing which one deserves to exist is harder. Relationships matter because important work still moves through credibility, empathy and shared commitment. Local context matters because a model may know a great deal about the world and still misunderstand how a Nigerian customer, regulator, employee or community will respond.

Your goal should therefore not be to compete with AI at producing the most predictable part of your job. Let the tool help with that. Your goal should be to move upward: from execution to orchestration, from receiving instructions to defining problems, from producing drafts to making decisions, and from completing tasks to owning outcomes.

This applies beyond the office. A person can use AI to understand a difficult subject, organise family finances, prepare questions for a professional, support a child’s learning or coordinate a community project. It should not replace qualified medical, legal or financial advice, and it should not make intimate decisions for us. But used with wisdom, it can increase our capacity to contribute to our families and communities.

The question is not only, “Will AI take my job?” A better set of questions is: Has AI changed what excellent performance looks like in my work? Is somebody else already producing more value with the same amount of time? Which part of my work has become cheaper? Which part has become more important? What must I learn now, before the market makes the answer obvious?

Do not wait until your employer announces a restructuring. Do not wait until a younger company serves your customers faster. Do not wait until the promotion goes to the colleague who quietly redesigned the work.

You do not need to know everything about AI today. You need to begin. Experiment responsibly. Measure what improves. Strengthen your judgment. Protect confidential information. Learn how to direct intelligent systems, and then use the capacity you gain to solve more important problems.

AI may never take your chair; it may never delete your job title. But if the standard of valuable work rises and you remain where you are, your value will still fall.

The good news is that the reverse is also true. You do not have to become somebody else to become more valuable. You can start with who you are, apply these tools with intelligence and integrity, and gradually become capable of work that was previously beyond your reach.

The real danger is not that AI becomes powerful. It is that AI becomes powerful while you remain unchanged.