What Experts Still Get Paid For When AI Makes Information Free

Assess whether AI can replace a particular consulting or coaching service by comparing its tasks, decisions and implementation requirements. My view that experience and judgement matter supplies a way to inspect the offer, not a forecast that either profession is safe.

Start with the work somebody is considering paying for. Identify the task, the decision it serves and what has to happen for that decision to become a working change. Those are more useful things to inspect than a prediction about an entire profession.

What is my view about readily available advice?

In my masterclass explanation at 02:58, I argue that people can get advice from AI that seems acceptable to them. My concern is that somebody outside a field may have difficulty judging what that advice leaves out.

At 03:35, I describe being pleased with a photograph I had taken on my phone and showing it to a photographer friend. The point of the story is the technical detail an experienced photographer can notice that I may miss.

That is my account and illustration. The photograph is not supplied here for assessment, and the story is not a study of AI quality, consultant performance or what buyers will pay.

By 04:48, I describe people deciding whom they trust and whose knowledge, history, tools and experience relate to their situation. That is my argument about expert value, not evidence that a particular category of work is protected from change.

Use the argument to inspect the offer. Which part helps the buyer recognise a problem with an apparently acceptable answer? What can they examine to decide whether your judgement adds something useful?

How do I separate an information task from a decision?

An information task might produce a summary, draft or list of options. Define its inputs and the evidence needed to check its output. Treat the output as a candidate until it meets those requirements.

A decision adds the circumstances that make one option fit: the budget, deadline, history, risk and promises already made. Record which condition changes the recommendation and what evidence supports that choice.

An AI-produced recommendation can be examined against those conditions too. The useful comparison is whether a specific output meets the job, not whether the person or tool is automatically entitled to be trusted.

Use this proposed comparison worksheet. It is an editorial aid for inspecting an offer, not a client case or a test I claim to have run.

Task: record the summary, draft or set of options being supplied. Check the source, accuracy, relevant omissions and the rule that defines completion.

Decision: record the recommendation for this situation. Check the constraints, reasons, alternatives and uncertainty that determine whether to accept it.

Implementation: record the change being put into use. Check authority, the working handoff, acceptance evidence, exceptions and responsibility for repair.

Record who performs each part and who reviews it. A tool may contribute at more than one stage. The record assigns checks and responsibility; it does not declare a permanent boundary around what software will be capable of.

What does implementation add beyond a plausible answer?

A recommendation still needs a route into the work. Identify the access, sequencing, review and changes required, including who can authorise them and how completion will be established.

Use the same standard for something a consultant supplies. A polished plan is not proof that the change was implemented. Record what was put into use and what remains unfinished.

Keep the exception visible. If the supplied information is incomplete, a dependency fails or the recommendation conflicts with another obligation, identify who decides what happens next.

For the broader division of work, use the AI acquisition handover check. This page examines what the service contains; that page checks a bounded delegation.

Where do experience and personality fit in that comparison?

In my AI-source explanation at 09:43, I describe personality, experience and the things someone has been through as material others may want to learn from. I argue for keeping the perspective of the person whose work is being examined.

That is my opinion about what makes the work worth encountering. The passage does not measure demand for a service or prove that personality prevents replacement.

Make the experience useful by explaining a choice, its source and its limit. The reader should be able to identify what you noticed and how it affected the decision, rather than merely learning an unusual fact about you.

Use content that sounds like you to preserve that source in the writing. For the relationship beyond one piece, see why trust does not go viral.

How can I show the difference without inventing a result?

Choose a piece of your own shareable work and put the task, decision and implementation record beside it. State what the artefact actually establishes and where the record stops.

For example, a supplied draft can show the output you reviewed. Your annotations can show the errors or missing context you identified. A later working change needs its own evidence; it cannot be inferred from the existence of the annotations.

That is an illustration of what to collect, not a report of my client work. Keep confidential material private and establish the applicable permission before using somebody else's project.

The proof-based content receipt handles the evidence behind an individual claim. Keep that record with the example instead of filling an empty outcome field with what probably happened.

What should I change in an offer built around available information?

List the promised deliverables and the decisions they serve. For each, record the source, required context, acceptance rule, implementation work and person responsible for exceptions. The no-sales-call acquisition system shows how visible judgement and useful implementation can help a buyer understand that value before a fit decision.

Then inspect what the buyer can already obtain and use adequately for that job. Test the actual output against the same requirements before claiming that your contribution or an AI tool is better.

Make the additional work explicit where it exists. Diagnosis, choosing between alternatives, adapting to constraints, implementing and repairing a change are concrete activities to explain and price. A label such as strategic advice still needs to show what someone receives.

If a proposed contribution cannot be supported, narrow the promise or improve the work. Adding a framework name or saying that you use AI does not supply the missing evidence.

What would justify a stronger claim about replacement?

Define the particular service, task conditions, quality threshold and observation period. A task comparison can then establish something about that task under those conditions. A claim about jobs or an entire market needs a different body of evidence.

Keep changes in demand, a decision to hire and an observed work outcome separate. My opinions and the comparison worksheet here supply no employment forecast, replacement probability or evidence of immunity for people who adopt AI.

The next useful step is to inspect one deliverable and the decision behind it. Show what the work requires, test the contribution and keep responsibility attached to the promises you make.

Get the daily email for more observations from my work.