𝐖𝐡𝐨 𝐜𝐡𝐨𝐬𝐞 𝐲𝐨𝐮𝐫 𝐀𝐈'𝐬 𝐯𝐚𝐥𝐮𝐞𝐬?

Who chose your AI’s values?

Most leadership teams have an answer ready for who bought their AI tools.

Far fewer can answer who shaped the behaviour those tools now produce.

The instinct is to look outward:

- the AI lab,
- the regulator,
- the technology provider.

But some of the most consequential choices happen much closer to home.

The picture hereby maps the eight points where an AI system’s values accumulate — and highlights the two points where your organisation is still holding the pen.

- 𝙏𝙝𝙚 𝙨𝙮𝙨𝙩𝙚𝙢 𝙥𝙧𝙤𝙢𝙥𝙩: the standing instruction your organisation writes and attaches to conversations with the model.
- 𝙏𝙝𝙚 𝙙𝙚𝙛𝙖𝙪𝙡𝙩: whichever model and settings appear when someone opens the tool without changing anything.

The interesting thing is that nobody usually makes these choices thinking:

“I am defining our organisation’s AI values.”

They are making reasonable operational decisions:

- choosing a tool,
- accepting a configuration,
- writing instructions,
- moving quickly.

But reasonable choices can accumulate into outcomes nobody explicitly designed.

Jared Spool found something similar years ago with Microsoft Word: fewer than 5% of users had changed a single default setting. Many assumed Microsoft had chosen those settings for a reason.

AI rollouts are likely to follow the same pattern.

The default becomes the de facto policy, because most people never become the person who changes what was already chosen.

Three questions I would ask any leadership team:

1. Who wrote our system prompt — and when was it last reviewed?
2. Which model and settings appear by default — and who decided that?
3. Which important decisions last quarter started with AI output and went through without meaningful challenge?

The answers are usually more revealing than the policy documents.

AI governance is not only about the layers we cannot control.

It is also about the layers we quietly own.

Swipe through, then ask:

Does anyone in your organisation actually know what your system prompt says?

#AIGovernance #Leadership #AIStrategy #ResponsibleAI

Contact

bruno.gentil@sherpaconsultingasia.com

© 2026. All rights reserved.