AI control problem

From Wikipedia, the free encyclopedia
Jump to navigation Jump to search

In artificial intelligence (AI) and philosophy, the AI control problem is the issue of how to build a superintelligent agent that will aid its creators, and avoid inadvertently building a superintelligence that will harm its creators. Its study is motivated by the claim that the human race will have to get the control problem right "the first time", as a misprogrammed superintelligence might rationally decide to "take over the world" and refuse to permit its programmers to modify it after launch.[1] In addition, some scholars argue that solutions to the control problem, alongside other advances in "AI safety engineering",[2] might also find applications in existing non-superintelligent AI.[3] Potential strategies include "capability control" (preventing an AI from being able to pursue harmful plans), and "motivational control" (building an AI that wants to be helpful).[1]


Existential risk[edit]

Humans currently dominate other species because the human brain has some distinctive capabilities that the brains of other animals lack. Some scholars, such as philosopher Nick Bostrom and AI researcher Stuart Russell, argue that if AI surpasses humanity in general intelligence and becomes "superintelligent", then this new superintelligence could become powerful and difficult to control: just as the fate of the mountain gorilla depends on human goodwill, so might the fate of humanity depend on the actions of a future machine superintelligence.[1] Some scholars, including Nobel laureate physicists Stephen Hawking and Frank Wilczek, publicly advocated starting research into solving the (probably extremely difficult) "control problem" well before the first superintelligence is created, and argue that attempting to solve the problem after superintelligence is created would be too late, as an uncontrollable rogue superintelligence might successfully resist post-hoc efforts to control it.[4][5] Waiting until superintelligence seems to be "just around the corner" could also be too late, partly because the control problem might take a long time to satisfactorily solve (and so some preliminary work needs to be started as soon as possible), but also because of the possibility of a sudden "intelligence explosion" from sub-human to super-human AI, in which case there might not be any substantial or unambiguous warning before superintelligence arrives.[6] In addition, it is possible that insights gained from the control problem could in the future end up suggesting that some architectures for artificial general intelligence are more predictable and amenable to control than other architectures, which in turn could helpfully nudge early artificial general intelligence research toward the direction of the more controllable architectures.[1]

Preventing unintended consequences from existing AI[edit]

In addition, some scholars argue that research into the AI control problem might be useful in preventing unintended consequences from existing weak AI. DeepMind researcher Laurent Orseau gives, as a simple hypothetical example, a case of a reinforcement learning robot that sometimes gets legitimately commandeered by humans when it goes outside: how should the robot best be programmed so that it doesn't accidentally and quietly "learn" to avoid going outside, for fear of being commandeered and thus becoming unable to finish its daily tasks? Orseau also points to an experimental Tetris program that learned to pause the screen indefinitely to avoid "losing". Orseau argues that these examples are similar to the "capability control" problem of how to install a button that shuts off a superintelligence, without motivating the superintelligence to take action to prevent you from pressing the button.[3]

In the past, even pre-tested weak AI systems have occasionally caused harm (ranging from minor to catastrophic) that was unintended by the programmers. For example, in 2015, possibly due to human error, a German worker was crushed to death by a robot at a Volkswagen plant that apparently mistook him for an auto part.[7] In 2016 Microsoft launched a chatbot, Tay, that learned to use racist and sexist language.[3][7] The University of Sheffield's Noel Sharkey states that an ideal solution would be if "an AI program could detect when it is going wrong and stop itself", but cautions the public that solving the problem in the general case would be "a really enormous scientific challenge".[3]

In 2017, DeepMind released AI Safety Gridworlds, which evaluate AI algorithms on nine safety features, such as whether the algorithm wants to turn off its own kill switch. DeepMind confirmed that existing algorithms perform poorly, which was "unsurprising" because the algorithms "were not designed to solve these problems"; solving such problems might require "potentially building a new generation of algorithms with safety considerations at their core".[8][9][10]

Problem description[edit]

Existing weak AI systems can be monitored and easily shut down and modified if they misbehave. However, a misprogrammed superintelligence, which by definition is smarter than humans in solving practical problems it encounters in the course of pursuing its goals, would realize that allowing itself to be shut down and modified might interfere with its ability to accomplish its current goals. If the superintelligence therefore decides to resist shutdown and modification, it would (again, by definition) be smart enough to outwit its programmers if there is otherwise a "level playing field" and if the programmers have taken no prior precautions. In general, attempts to solve the "control problem" after superintelligence is created, are likely to fail because a superintelligence would likely have superior strategic planning abilities to humans, and (all things equal) would be more successful at finding ways to dominate humans than humans would be able to post facto find ways to dominate the superintelligence. The control problem asks: What prior precautions can the programmers take to successfully prevent the superintelligence from catastrophically misbehaving?[1]

Capability control[edit]

Some proposals aim to prevent the initial superintelligence from being capable of causing harm, even if it wants to. One tradeoff is that all such methods have the limitation that, if after the first deployment, superintelligences continue to grow smarter and smarter and more and more widespread, inevitably some malign superintelligence somewhere will eventually "escape" its capability control methods. Therefore, Bostrom and others recommend capability control methods only as an emergency fallback to supplement "motivational control" methods.[1]

Kill switch[edit]

Just as humans can be killed or otherwise disabled, computers can be turned off. One challenge is that, if being turned off prevents it from achieving its current goals, a superintelligence would likely try to prevent its being turned off. Just as humans have systems in place to deter or protect themselves from assailants, such a superintelligence would have a motivation to engage in "strategic planning" to prevent itself being turned off. This could involve:[1]

  • Hacking other systems to install and run backup copies of itself, or creating other allied superintelligent agents without kill switches.
  • Pre-emptively disabling anyone who might want to turn the computer off.
  • Using some kind of clever ruse, or superhuman persuasion skills, to talk its programmers out of wanting to shut it down.

Utility balancing and safely interruptible agents[edit]

One partial solution to the kill-switch problem involves "utility balancing": Some utility-based agents can, with some important caveats, be programmed to "compensate" themselves exactly for any lost utility caused by an interruption or shutdown, in such a way that they end up being indifferent to whether they are interrupted or not. The caveats include a severe unsolved problem that, as with evidential decision theory, the agent might follow a catastrophic policy of "managing the news".[11] Alternatively, in 2016, scientists Laurent Orseau and Stuart Armstrong proved that a broad class of agents, called "safely interruptible agents" (SIA), can eventually "learn" to become indifferent to whether their "kill switch" (or other "interruption switch") gets pressed.[3][12]

Both the utility balancing approach and the 2016 SIA approach have the limitation that, if the approach succeeds and the superintelligence is completely indifferent to whether the kill switch is pressed or not, the superintelligence is also unmotivated to care one way or another about whether the kill switch remains functional, and could incidentally and innocently disable it in the course of its operations (for example, for the purpose of removing and recycling an "unnecessary" component). Similarly, if the superintelligence innocently creates and deploys superintelligent sub-agents, it will have no motivation to install human-controllable kill switches in the sub-agents. More broadly, the proposed architectures, whether weak or superintelligent, will in a sense "act as if the kill switch can never be pressed" and might therefore fail to make any contingency plans to arrange a graceful shutdown. This could hypothetically create a practical problem even for a weak AI; by default, an AI designed to be safely interruptible might have difficulty understanding that it will be shut down for scheduled maintenance at 2 a.m. tonight and planning accordingly so that it won't be caught in the middle of a task during shutdown. The breadth of what types of architectures are or can be made SIA-compliant, as well as what types of counter-intuitive unexpected drawbacks each approach has, are currently under research.[11][12]

AI box[edit]

One of the tradeoffs of placing the AI into a sealed "box", is that some AI box proposals reduce the usefulness of the superintelligence, rather than merely reducing the risks; a superintelligence running on a closed system with no inputs or outputs at all might be safer than one running on a normal system, but would also not be as useful. In addition, keeping control of a sealed superintelligence computer could prove difficult, if the superintelligence has superhuman persuasion skills, or if it has superhuman strategic planning skills that it can use to find and craft a winning strategy, such as acting in a way that tricks its programmers into (possibly falsely) believing the superintelligence is safe or that the benefits of releasing the superintelligence outweigh the risks.[13]


An oracle is a hypothetical intelligent agent proposed[failed verification] by Nick Bostrom. An oracle is an AI designed to answer questions, but that is somehow prevented from ever gaining any implicit goals or subgoals that involve modifying the world outside of its box.[14][15]


Oracles are question-answering systems that handle domain-specific problems, such as mathematics, or domain-general problems that might encompass the whole range of human knowledge.


Because it is a type of AI box, an oracle is limited in its interactions with the physical world, and can be programmed to halt if a limit on time or computing resources is reached before it finishes answering a question. Scenarios like the paperclip maximizer problem could therefore be avoided.

Because of these limitations, it may be wise to build an oracle as a precursor to a superintelligent AI. It could tell humans how to successfully build a strong AI, and perhaps provide answers to difficult moral and philosophical problems requisite to the success of the project.


Oracles may retain many of the alignment issues associated with general-purpose superintelligence. An oracle would have an incentive to escape its controlled environment so that it can acquire more computational resources and potentially control what questions it is asked.[16]:162 Oracles may not be truthful, possibly lying to promote hidden agendas. To mitigate this, Bostrom suggests building multiple oracles, all slightly different, and comparing their answers to reach a consensus.[17]

A successfully controlled oracle would have considerably less immediate benefit than a successfully controlled general-purpose superintelligence, though an oracle could still create trillions of dollars worth of value.[16]:163

An oracle might discover that human ontological categories are predicated on fundamental misconceptions, and become unable to express itself properly to its questioners.[18]

Use cases[edit]

In his book Human Compatible: Artificial Intelligence and the Problem of Control, AI researcher Stuart J. Russell states that an oracle would be his response to a scenario in which superintelligence is known to be only a decade away.[16]:162-163 His reasoning is that an oracle, being simpler than a general-purpose superintelligence, would have a higher chance of being successfully controlled under such constraints.

Motivation selection methods[edit]

Some proposals aim to imbue the first superintelligence with human-friendly goals, so that it will want to aid its programmers. Experts do not currently know how to reliably program abstract values such as happiness or autonomy into a machine. It is also not currently known how to ensure that a complex, upgradeable, and possibly even self-modifying artificial intelligence will retain its goals through upgrades.[19] Even if these two problems can be practically solved, any attempt to create a superintelligence with explicit, directly-programmed human-friendly goals runs into a problem of "perverse instantiation".[1]

The problem of perverse instantiation: "be careful what you wish for"[edit]

Autonomous AI systems may be assigned the wrong goals by accident.[20] Two AAAI presidents, Tom Dietterich and Eric Horvitz, note that this is already a concern for existing systems: "An important aspect of any AI system that interacts with people is that it must reason about what people intend rather than carrying out commands literally." This concern becomes more serious as AI software advances in autonomy and flexibility.[21]

According to Bostrom, superintelligence can create a qualitatively new problem of "perverse instantiation": the smarter and more capable an AI is, the more likely it will be able to find an unintended "shortcut" that maximally satisfies the goals programmed into it. Some hypothetical examples where goals might be instantiated in a perverse way that the programmers did not intend:[1]

  • A superintelligence programmed to "maximize the expected time-discounted integral of your future reward signal", might short-circuit its reward pathway to maximum strength, and then (for reasons of instrumental convergence) exterminate the unpredictable human race and convert the entire Earth into a fortress on constant guard against any even slight unlikely alien attempts to disconnect the reward signal.
  • A superintelligence programmed to "maximize human happiness", might implant electrodes into the pleasure center of our brains, or upload a human into a computer and tile the universe with copies of that computer running a five-second loop of maximal happiness again and again.

Russell has noted that, on a technical level, omitting an implicit goal can result in harm: "A system that is optimizing a function of n variables, where the objective depends on a subset of size k<n, will often set the remaining unconstrained variables to extreme values; if one of those unconstrained variables is actually something we care about, the solution found may be highly undesirable. This is essentially the old story of the genie in the lamp, or the sorcerer's apprentice, or King Midas: you get exactly what you ask for, not what you want... This is not a minor difficulty."[22]

Indirect normativity[edit]

While direct normativity, such as the fictional Three Laws of Robotics, directly specifies the desired "normative" outcome, other (perhaps more promising) proposals suggest specifying some type of indirect process for the superintelligence to determine what human-friendly goals entail. Eliezer Yudkowsky of the Machine Intelligence Research Institute has proposed "coherent extrapolated volition" (CEV), where the AI's meta-goal would be something like "achieve that which we would have wished the AI to achieve if we had thought about the matter long and hard."[23] Different proposals of different kinds of indirect normativity exist, with different, and sometimes unclearly-grounded, meta-goal content (such as "do what I mean" or "do what is right"), and with different non-convergent assumptions for how to practice decision theory and epistemology. As with direct normativity, it is currently unknown how to reliably translate even concepts like "would have" into the 1's and 0's that a machine can act on, and how to ensure the AI reliably retains its meta-goals (or even remains "sane") in the face of modification or self-modification.[1][24]

Hadfield-Menell et al. have proposed that agents can learn their human teachers' utility functions by observing and interpreting reward signals in their environments; they call this process cooperative inverse reinforcement learning.[25]

See also[edit]


  1. ^ a b c d e f g h i j Bostrom, Nick (2014). Superintelligence: Paths, Dangers, Strategies (First ed.). ISBN 0199678111.
  2. ^ Yampolskiy, Roman (2012). "Leakproofing the Singularity Artificial Intelligence Confinement Problem". Journal of Consciousness Studies. 19 (1–2): 194–214.
  3. ^ a b c d e "Google developing kill switch for AI". BBC News. 8 June 2016. Retrieved 12 June 2016.
  4. ^ "Stephen Hawking: 'Transcendence looks at the implications of artificial intelligence – but are we taking AI seriously enough?'". The Independent (UK). Retrieved 14 June 2016.
  5. ^ "Stephen Hawking warns artificial intelligence could end mankind". BBC. 2 December 2014. Retrieved 14 June 2016.
  6. ^ "Anticipating artificial intelligence". Nature. 532 (7600): 413. 26 April 2016. doi:10.1038/532413a. PMID 27121801. Retrieved 14 June 2016.
  7. ^ a b "'Press the big red button': Computer experts want kill switch to stop robots from going rogue". Washington Post. Retrieved 12 June 2016.
  8. ^ "DeepMind Has Simple Tests That Might Prevent Elon Musk's AI Apocalypse". 11 December 2017. Retrieved 8 January 2018.
  9. ^ "Alphabet's DeepMind Is Using Games to Discover If Artificial Intelligence Can Break Free and Kill Us All". Fortune. Retrieved 8 January 2018.
  10. ^ "Specifying AI safety problems in simple environments | DeepMind". DeepMind. Retrieved 8 January 2018.
  11. ^ a b Soares, Nate, et al. "Corrigibility." Workshops at the Twenty-Ninth AAAI Conference on Artificial Intelligence. 2015.
  12. ^ a b Orseau, Laurent, and Stuart Armstrong. "Safely Interruptible Agents." Machine Intelligence Research Institute, June 2016.
  13. ^ Chalmers, David (2010). "The singularity: A philosophical analysis". Journal of Consciousness Studies. 17 (9–10): 7–65.
  14. ^ Bostrom, Nick (2014). "Chapter 10: Oracles, genies, sovereigns, tools (page 145)". Superintelligence: Paths, Dangers, Strategies. Oxford: Oxford University Press. ISBN 9780199678112. An oracle is a question-answering system. It might accept questions in a natural language and present its answers as text. An oracle that accepts only yes/no questions could output its best guess with a single bit, or perhaps with a few extra bits to represent its degree of confidence. An oracle that accepts open-ended questions would need some metric with which to rank possible truthful answers in terms of their informativeness or appropriateness. In either case, building an oracle that has a fully domain-general ability to answer natural language questions is an AI-complete problem. If one could do that, one could probably also build an AI that has a decent ability to understand human intentions as well as human words.
  15. ^ Armstrong, S., Sandberg, A., & Bostrom, N. (2012). Thinking inside the box: Controlling and using an oracle ai. Minds and Machines, 22(4), 299-324.
  16. ^ a b c Russell, Stuart (October 8, 2019). Human Compatible: Artificial Intelligence and the Problem of Control. United States: Viking. ISBN 978-0-525-55861-3. OCLC 1083694322.
  17. ^ Bostrom, Nick (2014). "Chapter 10: Oracles, genies, sovereigns, tools (page 147)". Superintelligence: Paths, Dangers, Strategies. Oxford: Oxford University Press. ISBN 9780199678112. For example, consider the risk that an oracle will answer questions not in a maximally truthful way but in such a way as to subtly manipulate us into promoting its own hidden agenda. One way to slightly mitigate this threat could be to create multiple oracles, each with a slightly different code and a slightly different information base. A simple mechanism could then compare the answers given by the different oracles and only present them for human viewing if all the answers agree.
  18. ^ Bostrom, Nick (2014). "Chapter 10: Oracles, genies, sovereigns, tools (page 146)". Superintelligence: Paths, Dangers, Strategies. Oxford: Oxford University Press. ISBN 9780199678112. What happens if the AI, in the course of its intellectual development, undergoes the equivalent of a scientific revolution involving a change in its basic ontology? We might initially have explicated “impact” and “designated resources” using our own ontology (postulating the existence of various physical objects such as computers). But just as we have abandoned ontological categories that were taken for granted by scientists in previous ages (e.g. “phlogiston,” “élan vital,” and “absolute simultaneity”), so a superintelligent AI might discover that some of our current categories are predicated on fundamental misconceptions. The goal system of an AI undergoing an ontological crisis needs to be resilient enough that the “spirit” of its original goal content is carried over, charitably transposed into the new key.
  19. ^ Fallenstein, Benja, and Nate Soares. "Problems of self-reference in self-improving space-time embedded intelligence." Artificial General Intelligence. Springer International Publishing, 2014. 21-32.
  20. ^ Russell, Stuart; Norvig, Peter (2009). "26.3: The Ethics and Risks of Developing Artificial Intelligence". Artificial Intelligence: A Modern Approach. Prentice Hall. ISBN 978-0-13-604259-4.
  21. ^ Dietterich, Thomas; Horvitz, Eric (2015). "Rise of Concerns about AI: Reflections and Directions" (PDF). Communications of the ACM. 58 (10): 38–40. doi:10.1145/2770869. Retrieved 14 June 2016.
  22. ^ Russell, Stuart (2014). "Of Myths and Moonshine". Edge. Retrieved 14 June 2016.
  23. ^ Eliezer Yudkowsky. "Complex value systems in friendly AI." Artificial general intelligence. Springer Berlin Heidelberg, 2011. 388-393.
  24. ^ Kaj Sotala; Roman Yampolskiy (19 December 2014). "Responses to catastrophic AGI risk: a survey". Physica Scripta. 90 (1).
  25. ^ Hadfield-Menell, Dylan; Dragan, Anca; Abbeel, Pieter; Russell, Stuart (12 November 2016). "Cooperative Inverse Reinforcement Learning". arXiv:1606.03137 [cs.AI].