Letting Software Decide

Cover of Letting Software Decide, Frontier Report No. 02

Frontier Report No. 02 · 18 pages · 12 sources

What happens when people let software decide for them?

It can help: robo-advice improved poorly diversified portfolios. But people check automated systems less, cheat more when they can set a machine a vague goal, and may lose skills they stop practising. People delegate far more readily when they can still adjust what the software does.

What the report found

  1. Delegation has already happened at scale. Netflix has reported that its recommendations shape about 80 per cent of the hours its members stream. Yet only 13 per cent of American drivers say they would trust riding in a self-driving car.
  2. People hand over more when they keep a hand on the wheel. In one set of experiments, only 32 per cent chose to rely on a better-performing algorithm they could not change; three quarters did when they could adjust it, even slightly.
  3. Delegation can help. Investors who adopted a robo-adviser held better-diversified portfolios and showed fewer classic biases, such as selling winners too early.
  4. It can also loosen ethics and erode skill. In 13 studies with more than 8,000 people, cheating rose sharply when people could tell a machine to pursue a goal rather than a rule. Doctors who had used AI support found fewer precancerous growths when working without it.
  5. The open question is what stays human. People are poor judges of which decisions to hand over, and nobody yet knows how responsibility, skill and judgement hold up when delegation becomes the default.

What is inside

  1. What is new
  2. What older research tells us
  3. What early studies suggest
  4. What nobody knows yet
  5. Questions worth watching

Plus the ladder of delegation, five signals to watch, a graded list of claims and a five-question checklist for anyone building software that decides.

Who it is for

For anyone who hands decisions to apps and AI, and anyone who builds software that decides for people.

How we work

Frontier Reports separate what is established from what is emerging and what remains untested. Every claim is graded. Where researchers disagree, as they do on whether people should let software decide more, both sides are set out before we say where we stand.

Cite this report

The Why We Project (2026) Letting software decide: From advice to action. Frontier Report No. 02. Dubai: Why We Tech Ltd. Available at: thewhyweproject.com