arXiv subjects

Peter de Blanc

Publications and source records attributed to Peter de Blanc.

3 recordsLinked to original sources

arXiv2011

Ontological Crises in Artificial Agents' Value Systems

Decision-theoretic agents predict and evaluate the results of their actions using a model, or ontology, of their environment. An agent's goal, or utility function, may also be specified in terms of the states of, or entities within, its ontology. If the agent may upgrade or replace its ontology, it faces a crisis: the agent's original goal may not be well-defined with respect to its new ontology. This crisis must be resolved before the agent can make plans towards achieving its goals. We discuss in this paper which sorts of agents will undergo ontological crises and why we may want to create such agents. We present some concrete examples, and argue that a well-defined procedure for resolving ontological crises is needed. We point to some possible approaches to solving this problem, and evaluate these methods on our examples.

cs.AI↗

arXiv2009

Convergence of Expected Utility for Universal AI

We consider a sequence of repeated interactions between an agent and an environment. Uncertainty about the environment is captured by a probability distribution over a space of hypotheses, which includes all computable functions. Given a utility function, we can evaluate the expected utility of any computational policy for interaction with the environment. After making some plausible assumptions (and maybe one not-so-plausible assumption), we show that if the utility function is unbounded, then the expected utility of any policy is undefined.

cs.AI↗

arXiv2007

Convergence of Expected Utilities with Algorithmic Probability Distributions

We consider an agent interacting with an unknown environment. The environment is a function which maps natural numbers to natural numbers; the agent's set of hypotheses about the environment contains all such functions which are computable and compatible with a finite set of known input-output pairs, and the agent assigns a positive probability to each such hypothesis. We do not require that this probability distribution be computable, but it must be bounded below by a positive computable function. The agent has a utility function on outputs from the environment. We show that if this utility function is bounded below in absolute value by an unbounded computable function, then the expected utility of any input is undefined. This implies that a computable utility function will have convergent expected utilities iff that function is bounded.

cs.AI↗