Intelligent Agent Foundations Forumsign up / log in
Resolving human inconsistency in a simple model
post by Stuart Armstrong 73 days ago | Abram Demski likes this | 1 comment

A putative new idea for AI control; index here.

This post will present a simple model of an inconsistent human, and ponder how to resolve their inconsistency.

Let \(\bf{H}\) be our agent, in a turn-based world. Let \(R^l\) and \(R^s\) be two simple reward functions at each turn. The reward \(R^l\) is thought of as being a ‘long-term’ reward, while \(R^s\) is a short-term one.


Define \(R^l_t\) as the agent’s \(R^l\) reward at turn \(t\) (and similarly \(R^s_t\) for \(R^s\)). Then, at turn \(t\), the agent \(\bf{H}\) has reward:

  • \(R_t=\sum_{\tau=0}^\infty (\gamma_l)^\tau R^l_{t+\tau}+ (\gamma_s)^\tau R^s_{t+\tau}\),

with constants \(0<\gamma_s < \gamma_l \leq 1\).

Essentially the \(R_s\) and the \(R_l\) have different discount rates, with the reward from \(R_s\) fading much faster than than that of \(R_l\). Therefore the agent will be motivated to get the \(R_s\) reward, but only if they can get this in the short-term. Sex, drugs, food, and many other pleasures often have these features (though they are, of course, much more complicated).

The inconsistency is that the human will continually reset their \(R_t\) at each turn. If there were a single discount rate, that wouldn’t a problem, as that would just scale the whole reward function, and reward functions, like utility functions, give the same decisions when scaled.

But with two discount rates, this is inconsistent. The agent will try and follow \(R_l\) for long-term planning, but this will be disrupted if they encounter an \(R_s\) along the way (and then presumably berate themselves for the lack of self-discipline). This can also be seen as a variant of the “humans are composed of multiple subagents” model, with \(R_s\) corresponding to short-term greedy subagent.

So, we have a simple and not-completely-implausible model of an inconsistent human. The question is, how do we resolve it? None of the obvious approaches are ideal, but it’s worth looking at their features.

Freeze the reward

This is the most obvious approach: simply freeze the reward, so that the reward at time \(t'>t\) is simply the same as reward at time \(t\) (though, in the absence of time-travel, the rewards between \(t\) and \(t'\) are no longer relevant).

This is the obvious approach; in practice, though, it will become equivalent with simply forgetting about \(R_s\) entirely. After a few turns, the exponential shrinkage of the factor \((\gamma_l/\gamma_s)^\tau\) will make \(R_s\)’s typical contribution insignificant. So this approach involves destroying one of \(\bf{H}\)’s sources of reward almost entirely.

Balance the rewards

Another approach would be to balance the rewards, set \(\gamma_l=\gamma_s\), either at the initial value of \(\gamma_l\), the initial value of \(\gamma_s\), or some other value.

This would make the the reward consistent, but has the opposite problem as the previous approach: the long-term importance of \(R_s\) is now massively magnified relative to \(R_l\), so now the long-term plans will prioritise \(R_s\) above \(R_l\) much more than before.

Narrative/frequentist/feature based approach

Since we’re not supposed to do this, let’s anthropomorphise \(\bf{H}\). We can imagine that \(\bf{H}\) has some sort of narrative about their existence – they see themselves as being a certain type of person (possibly mainly connected with \(R_l\)), who has some quirks/indulgences/sins (possibly mainly connected with \(R_s\)).

If they want to extirpate \(R_s\) entirely (“sin”), this is the same as the “Freeze the reward” approach. But they may instead prefer to live their life with roughly the same proportion of \(R_s\) as before (“quirk”), or slightly less (“indulgence”).

In that case, the new \(R\) would be chosen for consequentialist reasons. Not by looking at the individual terms \(R_l\), \(R_s\), \(\gamma_l\), and \(\gamma_s\), but by looking at the consequences of following \(R_t\) under “typical” circumstances, and designing the new reward to replicate this behaviour (and this distribution of reward features), while allowing more efficiency. This is, in itself, an interesting IRL problem. But it seems to make sense for humans, as we define ourselves a lot by what we do and experience, rather than by the pleasures and choices that lead up to those experiences.



by Vadim Kosoy 25 days ago | link

Freezing the reward seems like the correct answer by definition, since if I am an agent following the utility function \(R\) and I have to design a new agent now, then it is rational for me to design the new agent to follow the utility function I am following now (i.e. this action is usually rated as the best according to my current utility function).

reply



NEW LINKS

NEW POSTS

NEW DISCUSSION POSTS

RECENT COMMENTS

This is exactly the sort of
by Stuart Armstrong on Being legible to other agents by committing to usi... | 0 likes

When considering an embedder
by Jack Gallagher on Where does ADT Go Wrong? | 0 likes

The differences between this
by Abram Demski on Policy Selection Solves Most Problems | 0 likes

Looking "at the very
by Abram Demski on Policy Selection Solves Most Problems | 0 likes

Without reading closely, this
by Paul Christiano on Policy Selection Solves Most Problems | 1 like

>policy selection converges
by Stuart Armstrong on Policy Selection Solves Most Problems | 0 likes

Indeed there is some kind of
by Vadim Kosoy on Catastrophe Mitigation Using DRL | 0 likes

Very nice. I wonder whether
by Vadim Kosoy on Hyperreal Brouwer | 0 likes

Freezing the reward seems
by Vadim Kosoy on Resolving human inconsistency in a simple model | 0 likes

Unfortunately, it's not just
by Vadim Kosoy on Catastrophe Mitigation Using DRL | 0 likes

>We can solve the problem in
by Wei Dai on The Happy Dance Problem | 1 like

Maybe it's just my browser,
by Gordon Worley III on Catastrophe Mitigation Using DRL | 2 likes

At present, I think the main
by Abram Demski on Looking for Recommendations RE UDT vs. bounded com... | 0 likes

In the first round I'm
by Paul Christiano on Funding opportunity for AI alignment research | 0 likes

Fine with it being shared
by Paul Christiano on Funding opportunity for AI alignment research | 0 likes

RSS

Privacy & Terms