Self-awareness is often sold as a private act: sit quietly, look inward, know yourself. That is useful, but too small. In real decisions, self-awareness works more like system monitoring: it helps people and teams detect stress, bias, uncertainty and drift before a decision hardens.

The problem hiding in plain sight

We like to treat decisions as clean intellectual events.

Gather the data. Compare the options. Pick the answer.

That is the tidy version. It is also the version that makes us feel most grown-up. But most important decisions do not arrive in tidy conditions. They arrive when people are tired, incentives are tangled, time is short, confidence is uneven and the room has already started leaning towards an answer.

This is where self-awareness matters.

Not the soft-focus version. Not the one that says, “I am the kind of person who likes clarity,” which is usually just a decorative way of saying “I dislike ambiguity”. I mean self-awareness as monitoring.

What is happening in me?

What is happening in the group?

What are we pretending not to know?

What is the system pushing us to decide before we have understood the cost?

Cai’s life-stage framing gives us a useful human starting point: self-awareness is not static. It changes as people move through development, culture and responsibility. The current research signal adds something sharper. Self-awareness is partly embodied, partly cultural, and now increasingly relevant to AI-assisted decisions.

Interoception — the ability to sense and interpret internal bodily signals — is now being studied as part of emotional regulation, self-regulation and embodied consciousness. Recent work links interoceptive awareness with adaptive functioning and time perspective, while wider reviews describe interoception as central to how people regulate emotion and make sense of experience. (PMC)

That sounds academic until you have sat in a delivery meeting where everyone knows the plan is unrealistic, but nobody wants to be the difficult one.

The trade-off is not reflection versus action.

It is speed versus understanding.

Why this matters now

AI has made it easier to produce a plausible answer quickly.

That is useful. I use these tools. I am not interested in pretending that the old way was purer. Across mainframe, client-server, cloud, agile and now AI, every platform shift has reduced some form of friction. That is usually the point.

But not all friction is waste.

Some friction is how judgement forms.

A 2025 paper on AI-supported personal decision-making argues that large language models can steer people towards solutions while reducing their metacognitive awareness of how they are thinking before they decide. The authors call attention to “pre-decision reflection” — the messy but important work that happens before an answer appears. (arXiv)

That is the bit organisations are at risk of removing.

Not deliberately. Nobody is writing a strategy called “Reduce Human Judgement by Accident”. But when the operating rhythm rewards fast summaries, clean options papers and rapid alignment, the organisation starts selecting for fluency over awareness.

It looks productive.

It may be making the decision system less resilient.

The wrong way to frame it

“We just need people to be more self-aware.”

True, but weak.

It makes self-awareness sound like a personality upgrade. As though the problem is that individuals have failed to become sufficiently wise in their own time.

That lets the system off the hook.

If every governance forum rewards confidence, punishes uncertainty and treats challenge as delay, then asking people to “be more self-aware” is mostly theatre.

“AI will help us make better decisions because it gives us more information.”

Sometimes it will.

More information can improve a decision, especially when the current process is slow, patchy or biased towards whoever speaks loudest. But information is not judgement. A better summary does not automatically create a better decision. It can just make a premature decision look more respectable.

There is also a cultural problem. Cross-cultural research on emotional self-awareness suggests that self-awareness does not present the same way in every context; different tasks can reveal different patterns between UK and Japanese participants. (Springer)

So even the instruction “speak up if something feels wrong” carries assumptions. In one culture, direct challenge may be treated as responsible. In another, it may be socially costly. In one team, doubt is valued. In another, doubt is quietly filed under “not commercial enough”.

Self-awareness is not just an individual virtue.

It is an operating condition.

A better framework: the Decision Monitoring Loop

I would not build this as a motivational poster. I would build it like a lightweight control loop.

1. Body signal: what is the human system telling us?

Before the decision, notice the physical and emotional state of the people making it.

This is not mystical. It is basic operational awareness. Fatigue, defensiveness, urgency and fear all change decision quality. Research on embodied decision-making argues that decisions and actions are not separate events; they interact dynamically through feedback loops. (PLOS)

Example: A team approves a date because everyone is exhausted by the argument, not because the date is credible.

Question: What are we feeling that may be masquerading as logic?

2. Context signal: what is the environment rewarding?

Look at the incentives around the decision.

Are people being rewarded for surfacing risk, or for keeping momentum? Is the meeting designed to understand the decision, or to protect a previous commitment?

Example: A programme board asks for “confidence”, but every honest caveat is treated as a delivery problem.

Question: What answer does this environment quietly want?

3. Culture signal: who can safely say what?

Self-awareness changes shape across cultures, roles and power structures. A senior leader saying “challenge me” does not automatically create psychological safety.

Example: A supplier knows the plan has a flaw, but the commercial relationship makes direct challenge feel expensive.

Question: Who would pay the highest price for telling the truth here?

4. AI signal: what has been made too easy?

AI can help people explore options. It can also collapse the useful discomfort that comes before a decision. Emerging work on AI-assisted critical thinking is exploring systems that prompt people to examine their own rationales, not just consume machine-generated recommendations. (arXiv)

Example: A project manager asks AI for a risk response and accepts the most polished answer, without first naming the actual trade-off.

Question: What thinking did the tool skip on our behalf?

5. Drift signal: what has changed since we first believed this?

Every decision has a half-life.

The trouble is that people attach emotionally to earlier assumptions. Once a plan has been socialised, costed and presented, noticing drift starts to feel like disloyalty.

Example: The test plan was sensible six weeks ago. Since then, dependencies moved, defects changed and team availability shifted. The document still looks official. The system has moved.

Question: What would make this decision wrong now, even if it was right before?

What to watch

Watch for these signals. They are small, but they matter.

1. Fast agreement after a difficult question
   The room may be aligning to escape discomfort, not because the issue is resolved.
2. A polished AI answer with no visible reasoning trail
   The output may be good. The missing part is ownership.
3. Risk language that becomes decorative
   If every risk is “being managed”, but nothing changes, the risk process has become wallpaper.
4. People saying “we all know”
   Often means the assumption has stopped being tested.
5. Challenge only coming from the same few people
   That is not a healthy challenge culture. That is dependency on a small number of organisational immune cells.
6. Confidence rising while evidence stays flat
   This is one of the clearest signs that the system is managing anxiety rather than uncertainty.

What to do next

1. Add a pre-decision pause

What to do: Before major decisions, ask everyone to write down the trade-off in one sentence before discussion starts.

Why it matters: It prevents the loudest framing from becoming the default framing.

Question: What are we choosing between, really?

2. Separate option generation from option selection

What to do: Use AI and group discussion to widen options first. Decide later.

Why it matters: When generation and selection happen together, people fall in love with the first workable answer.

Question: Have we created alternatives, or just refined the first answer?

3. Make uncertainty visible

What to do: Ask for confidence levels, assumptions and evidence quality, not just recommendations.

Why it matters: A decision can be right with low confidence, or wrong with beautiful formatting.

Question: What do we know, what do we infer, and what are we hoping?

4. Design challenge for culture and power

What to do: Use anonymous pre-reads, structured dissent or role-based challenge where direct disagreement is difficult.

Why it matters: “Speak freely” is not a mechanism. It is a wish.

Question: How would someone raise a concern here without losing face or influence?

5. Revisit decisions when the system changes

What to do: Put review triggers into decisions: date, dependency, defect level, cost movement, customer impact or regulatory change.

Why it matters: A decision without a review trigger becomes a belief.

Question: What signal would tell us this decision needs reopening?

Closing thought

The freeing part is that self-awareness does not require everyone to become endlessly introspective.

That would be unbearable. Also slow.

The better aim is more practical: build decision environments that can sense themselves. People should be able to notice when they are tired, cornered, overconfident or quietly conforming. Teams should be able to notice when speed is being confused with clarity. AI should help expose the reasoning, not just polish the recommendation.

The question is not whether we can decide faster.

We already can.

The question is whether we can still detect when fast has become fragile.

Sources used

• Tarvirdians, M., Chandrasegaran, S., Hung, H., Jonker, C. M. and Oertel, C. — “Reflection Before Action: Designing a Framework for Quantifying Thought Patterns for Increased Self-awareness in Personal Decision Making”, arXiv, 2025.
• Klamut, O. — “Bridging interoception and time perspective: toward an embodied model of consciousness”, Frontiers in Psychology, 2026.
• Lazzarelli, A. et al. — “Interoceptive Ability and Emotion Regulation in Mind–Body Interventions”, 2024.
• Priorelli, M. et al. — “Embodied decisions as active inference”, PLOS Computational Biology, 2025.
• Huggins, C. F. et al. — “Cross-cultural differences in self-reported and behavioural measures of emotional self-awareness”, BMC Research Notes, 2023.
• Li, S. et al. — “Understanding the Effects of AI-Assisted Critical Thinking on Human Decision-Making”, arXiv, 2026.