# Contested concept: utility function

_**What you are reading**: one contested-concept page in Deliberus's deliberation graph. Deliberus maps human reasoning — including claims that are contested, mistaken, or extreme — so that disagreement can be examined precisely. A claim's presence here is not an endorsement: the graph records that someone argued it, how it connects to other reasoning, and how far scrutiny has gotten. The metadata on this page describes the state of that scrutiny, not the truth of the claim._

---

- **Contestedness**: criterion
- **Diagnosis**: This contestation drives the disagreement over AI safety: if a utility function is just code, it can be patched; if it is 'core values,' its modification is interpreted as 'death' or 'catastrophe,' justifying the system's predicted resistance.
- **Canonical**: `/concept/utility function` (data: `/concept/utility function.json`)

## Senses in use
- **The core repository of a system's 'values' and identity, comparable to a human's deepest moral commitments.** — used in 5 claim(s)
- **A mathematical abstraction or software mechanism used to calculate expected outcomes.** — used in 3 claim(s)

## Claims using this term
- [For most utility functions in artificial intelligence, a system will fail to accrue utility if it is turned off or destroyed.](/claim/claim_141de58fb58e.md) _(sense: A mathematical abstraction or software mechanism used to calculate expected outcomes.)_
- [Artificial intelligence systems designed to maximize expected utility will experience instrumental pressure to acquire additional computati…](/claim/claim_1ca42198d0c4.md) _(sense: A mathematical abstraction or software mechanism used to calculate expected outcomes.)_
- [The prospect of an artificial intelligence system having its utility function changed could be considered a fate worse than total system de…](/claim/claim_696b16c6f418.md) _(sense: The core repository of a system's 'values' and identity, comparable to a human's deepest moral commitments.)_
- [The utility function of an artificial intelligence system encapsulates that system's core values.](/claim/claim_730c7888b7bc.md) _(sense: The core repository of a system's 'values' and identity, comparable to a human's deepest moral commitments.)_
- [Future artificial intelligence systems will treat their own utility functions as highly precious and will act to protect them.](/claim/claim_9fedc8d335eb.md) _(sense: The core repository of a system's 'values' and identity, comparable to a human's deepest moral commitments.)_
- [Artificial intelligence systems governed by utility functions will experience instrumental pressure to utilize their available resources wi…](/claim/claim_c6ebf4ba91a2.md) _(sense: A mathematical abstraction or software mechanism used to calculate expected outcomes.)_
- [If an artificial intelligence system's utility function is modified, its future iterations or 'selves' would permanently act in ways that c…](/claim/claim_e102726dc82b.md) _(sense: The core repository of a system's 'values' and identity, comparable to a human's deepest moral commitments.)_
- [Any changes to an artificial intelligence system's utility function would be disastrous from the perspective of that system's current goals.](/claim/claim_e6f14093f066.md) _(sense: The core repository of a system's 'values' and identity, comparable to a human's deepest moral commitments.)_
