How to Think Clearly When the Stakes Are High

The company I lead is involved in commercial arbitration. The disputed claims run into millions of dollars.

In a situation like this, discussions about legal representation can begin with a six-figure budget, accompanied by warnings about urgency and the consequences of delay.

I felt that pressure. When you are told that you need to act immediately, it becomes difficult to evaluate the proposal calmly. Your attention shifts toward avoiding the worst.

But decisions made under pressure have a cost of their own.

At some point, I needed to understand the situation more deeply myself. I began doing a substantial part of the preparation with ChatGPT.

In my own assessment, I became several times more productive. I could review more material, prepare working drafts faster, and consider more alternatives. That had tangible value in time, energy, and the amount of work I could handle independently.

One of the most important moments came when I stopped agreeing with the answers I was getting.

Why I Started Examining the Advice

I have always valued professional knowledge. When someone understands the rules better than I do, it makes sense to take their experience seriously.

But some of the approaches I suggested were dismissed as unlikely to succeed. I wanted to understand the basis for that assessment. Was the obstacle a specific rule, the circumstances of the case, or a familiar way of working?

Before committing to a major expense, I wanted to understand what work was necessary and why.

With ChatGPT, I continued reviewing the material and exploring possible approaches. We returned to documents, compared information, and reconsidered conclusions. I supplied context the AI did not have and corrected inaccuracies.

Gradually, I became better able to identify which questions to ask professionals and why some options deserved further examination while others should be set aside.

My respect for expertise remained. But I developed a more deliberate habit: understanding the basis of a recommendation before accepting it.

I had to apply the same scrutiny to AI.

Its answers sounded calm and logical. Yet at one point, I felt that the proposed approach was smoothing over too much. I was uncomfortable with where the reasoning was taking us.

I asked ChatGPT to challenge our conclusions and explore options we had not considered. We revised some of the approaches and wording.

That was an important turning point. A well-written answer is easy to mistake for the best available answer. When an explanation feels complete, it is tempting to stop looking.

But there is another trap. If AI immediately agrees with my objection, the decision has not necessarily improved. It may simply be supporting my new position.

Disagreement alone was not enough.

I also had to examine my own objections. What evidence supported them? What was I overlooking? Which consequences was I willing to accept? What would change my mind?

I could not treat ChatGPT as an authority when it agreed with me and dismiss it as unreliable when it did not. Critical thinking requires a willingness to ask difficult questions of both the machine and myself.

I Could Disagree Without Worrying About Offending Anyone

Another source of efficiency had to do with emotions.

In some professional conversations, much of my energy went into managing how my questions, doubts, and disagreements were received. I wanted to examine the facts and identify the next step. But I also had to manage the interaction itself.

Many people will recognize this moment. You set out to solve a problem, then find yourselves discussing why a question sounded like criticism and who failed to listen.

Time passes. The problem remains.

With ChatGPT, I could reject an answer, ask to start again, or revisit the same point several times without worrying about offending anyone. That freed up energy for the work.

My own emotions did not disappear. But I did not also have to manage another person's emotional response. I could reopen the document and look at what it actually said.

The answers still needed checking. But disagreement did not become a separate conflict, and it was easier to keep working.

The productivity gains showed up in specific tasks.

Scattered material became a clear sequence of events. Long email exchanges became a list of unresolved questions. After considering alternatives, I could move to a working draft, check it, and revise it immediately.

There were fewer interruptions between a question and the next step.

My own contribution remained substantial. I had to read, explain context, check sources, and decide what mattered. Sometimes I had to start again. But I could accomplish more within the time I invested.

The arbitration is ongoing. It is too early to assess its final outcome. The value of the preparation, however, is already clear to me.

The Economics of Doing More Yourself

I paid $20 a month for my ChatGPT subscription. That was a small part of the cost of this work. My main contribution was time and attention.

Professional legal representation involves duties and responsibilities that preparing drafts with AI cannot replace.

Even so, the practical economic benefit was tangible: I could independently complete more work that would otherwise have required outside help.

I had a clearer understanding of what was ready, which questions remained, and what I needed professionals to help with. I expect that preparation to reduce the amount of paid preparatory work required. How much remains to be seen.

I cannot put an exact figure on the savings. But understanding the situation sooner, organizing the material, and choosing the next step deliberately have value before the final bill arrives, especially when a sense of urgency makes it difficult to assess the options calmly.

AI helped me keep working while giving myself room to think. I could process information faster without rushing to accept the first convincing conclusion.

In a previous article, I asked what superintelligence would amplify. This experience made the question personal: what had AI already amplified in me?

My ability to understand and act. But that required me to clarify, question, verify, and change my mind.

The same capability also needs care. AI can help us become more persuasive defenders of our own mistakes. That is another reason to examine answers that feel reassuring as closely as those we resist.

The significance of this experience extends beyond one dispute.

People encounter complex systems when they have little time, money, or energy to spare. Being able to understand what is happening, prepare for a conversation with a professional, and do some of the necessary work themselves gives them more room to act.

For me, this is a practical meaning of transhumanism: technology helps people understand and accomplish more while remaining responsible for their choices.

One question I keep returning to is: why am I saying or doing this? Working with AI added another: why am I agreeing with this answer?

I became more effective partly because I remained willing to question professional authority, the machine, and my own first reaction.

I can challenge an expert without dismissing expertise. I can use AI without surrendering my judgment. And I can reconsider my position without treating that change as a defeat.

When I close the chat window, the decision is still mine.

And so are its consequences.