Not necessarily an order

A public quarrel can be a curious kind of advertisement.

No one has to ask us to buy anything. It may be enough to get us caught up in the question, provoke us, or make us want to choose a side. The company names come along for the ride.

The Norwegian newspaper Morgenbladet’s coverage of Microsoft AI chief Mustafa Suleyman’s criticism of Anthropic is a concrete example of the discussion moving beyond the companies’ own writings into editorial coverage in Norway. The article was published on 30 September 2026. That demonstrates attention, not whatever commercial benefit anyone might derive from it. [1]

In the post that prompted the coverage, dated 16 September 2026, Suleyman argues against treating AI as conscious beings and against training models to behave as though they have feelings and moral claims of their own. At the same time, he presents Microsoft’s own approach as an alternative. Criticism of a competitor and the presentation of his own project appear in the same text. [2]

Two days earlier, Microsoft had also opened a public consultation on its Humanist AI Code of Conduct. The debate therefore also concerns competing companies’ proposals for how future AI should be developed. [3]

That does not invalidate the arguments. A commercial interest is not a refutation.

But it gives us reason to ask another question: when a company leader discusses humanity’s future, where does the substantive contribution end and the positioning begin?

Perhaps there is not always a clear dividing line.

The disagreement may be sincere. The concern may be serious. Media coverage may be critical and uncomfortable for those it discusses. Even so, the discussion may give companies visibility and establish their leaders as central interpreters of the technology they themselves are developing.

That is a possible publicity benefit. It is not the same as a documented, staged advertising campaign.

The distinction is worth preserving.

Before I have finished reading

I notice how tempting the simple explanation is.

This is just advertising.

It is a satisfying thought. Suddenly I have seen through something. The others are discussing consciousness; I have spotted the business model.

But perhaps this judgment came a little too quickly as well.

Have I examined the arguments? Or have I found an explanation that means I no longer have to?

Anthropic’s actual position, for instance, is not a straightforward declaration that Claude is conscious. The company’s foundational document expresses uncertainty about the model’s possible consciousness and moral status. It also describes tensions between the company’s commercial strategy and the qualities it wants to develop in the model. [4]

The same document requires Claude not to undermine legitimate human control, and not to use lying, sabotage or other illegitimate means to resist correction or shutdown. That part, too, belongs in any assessment of the criticism. [4]

I can still disagree. But now I have to disagree with something that is actually there.

That is less comfortable work than choosing a side.

And perhaps the essay’s real question begins right here:

What is it in me that is so eager to be done with the assessment?

My own programs

The question of whether a machine has subjective experience is not the same as whether a person notices their own reactions. One does not settle the other.

Still, the discussion can serve as a mirror.

Picture a small encounter.

Someone says: “We’ve tried that before.”

The words are barely out before a reply is on its way.

You never listen.

Perhaps it was a dismissal. Perhaps it was useful information. Perhaps the person was tired, or trying to save us from a mistake.

But before I even ask, I may already have assigned an intention to the sentence, put the person in a category and found my own role in the situation.

I heard a few words.

The rest of the story arrived very quickly.

“Program” can be a useful image for this. Not as a claim that a human being is a computer, but as a name for something rehearsed that begins to run: an expected sequence, a familiar defence, an old explanation keen to be used again.

Getting rid of all such things would be a strange kind of liberation.

I do not want to learn to ride a bicycle again every morning. I want to keep the words, the craftsmanship, the music, and what my body knows without my having to explain every step to it.

The question is not whether I should have learned patterns.

The question is whether I can sometimes notice when one of them has begun to determine more than the situation warrants.

An experience can be useful without being the whole of reality.

A small interval

Someone says something.

My chest tightens.

I do not always know why. Nor do I need to turn the feeling immediately into an explanation about the nervous system, childhood or the other person’s personality.

At first there is just this:

The words I heard. The tightening I noticed. The reply beginning to take shape.

Could they remain three different things for a moment?

Perhaps it is the joining together that happens so fast. The words, the feeling and the interpretation become a single event: He is attacking me.

That may be right. I may have understood the situation well.

But it may also be something I should examine before letting the ready-made response take over.

A small interval is no guarantee of wisdom. I can be wrong after a long time thinking, too.

Still, there is a difference between having a reaction and noticing that I am having it.

Then at least there is a question where, a moment earlier, there was only an order.

My own prompts, too

“You ought to.”

“Everyone knows.”

“That’s how things are done.”

It is tempting to call such sentences interpersonal prompt injection: small attempts to get particular instructions to steer the next response.

As an image, it can be illuminating. But it can also become far too convenient.

For I have prompts of my own.

I have to explain myself.

I have to show that I have understood more.

This is just advertising.

And perhaps the most cunning one:

I am observing. The others are just following their programs.

There it was again. Another role to step into. The discerning spectator, neatly positioned outside the thing he sees through.

But I am not outside it.

With funding, I would develop an AI company myself. I have the capacity to do that. So I am not merely a spectator in this field. I have ambitions within it.

That does not make the observation about the media debate less relevant. But it also applies to a position I could come to occupy myself. If one day I were presenting my own company, my professional convictions would stand alongside a wish for people to listen, to trust the work and to support it.

It would then be reasonable to ask me the same question: what is documented, what is my conviction, and what am I hoping to bring others on board with?

This text, too, is trying to hold your attention.

The difference therefore cannot simply be that I think something different from the company leaders. It must also lie in what the text lets you examine — and whether it gives you room to discover something other than what I had in mind.

Outside

Yellow, green and orange leaves lie on the forest floor.

The wind moves some of them. A cool autumn breeze touches my face.

I can immediately begin turning the scene into an explanation. Nature as the opposite of technology. Stillness as an answer to noise. The forest as a place where a human being is finally authentic.

It could make a beautiful paragraph.

But perhaps, right now, it is enough to look at the leaves.

An uneven edge. A darker patch. Something moving while something else stays where it is.

The thought of the discussion returns. I do not have to chase it away. It can remain a thought, without the whole forest having to become its illustration.

Later I may need to take a position. Say no. Write a reply. Examine a claim more closely.

Letting an interpretation wait is not the same as giving up judgment.

Perhaps, sometimes, this is how judgment gets slightly better working conditions.

We began with the question of whether a machine can be more than what it has been trained to say.

That question has not been settled here.

But along the way another, closer question emerged:

When my reply is already on its way — can I notice it before I turn it into truth?

A thought has arisen.

It may be good. It may be old. It may need to be tested against something I have not yet seen.

But it is not necessarily an order.