Error-Correction for AI Safety

Aliman, Nadisha-Marie; Elands, Pieter; Hürst, Wolfgang; Kester, Leon; Þórisson, Kristinn R.; Werkhoven, Peter; Yampolskiy, Roman V.; Ziesche, Soenke

doi:10.1007/978-3-030-52152-3_2

Cited by 7 publications

(21 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…One of the strengths of Definition 1 is that it is uniform across different AGIs: many different AGIs might internally operate based on different definitions of knowledge, but Definition 1 works equally well for all these different AGIs regardless of those different internal knowledge definitions 7 . We can contrast this with difficulties that could arise from a more experimental approach.…”

Section: An Elegant Definition Of Mathematical Knowledgementioning

confidence: 99%

“…At the same time, the problem is often brushed aside as something only philosophers care about: pragmatists rarely spend time on this sort of debate. One exception is in the study of agents with Artificial General Intelligence (AGIs, or Type II AIs in the terminology of [7]), where even the staunchest pragmatists admit the importance of the question.…”

Section: Introductionmentioning

confidence: 99%

“…This is reminiscent of Williamson's contextualism[23] 7. This is reminiscent of Elton's proposal that instead of trying to interpret an AI's outputs by focusing on specific low-level details of a neural network, we should instead let the AI explain itself[12].…”

mentioning

confidence: 99%

See 2 more Smart Citations

Short-Circuiting the Definition of Mathematical Knowledge for an Artificial General Intelligence

Alexander

2021

Software Engineering and Formal Methods. SEFM 2020 Collocated Workshops

View full text Add to dashboard Cite

We propose that, for the purpose of studying theoretical properties of the knowledge of an agent with Artificial General Intelligence (that is, the knowledge of an AGI), a pragmatic way to define such an agent's knowledge (restricted to the language of Epistemic Arithmetic, or EA) is as follows. We declare an AGI to know an EA-statement φ if and only if that AGI would include φ in the resulting enumeration if that AGI were commanded: "Enumerate all the EA-sentences which you know." This definition is non-circular because an AGI, being capable of practical English communication, is capable of understanding the everyday English word "know" independently of how any philosopher formally defines knowledge; we elaborate further on the non-circularity of this circular-looking definition. This elegantly solves the problem that different AGIs may have different internal knowledge definitions and yet we want to study knowledge of AGIs in general, without having to study different AGIs separately just because they have separate internal knowledge definitions. Finally, we suggest how this definition of AGI knowledge can be used as a bridge which could allow the AGI research community to import certain abstract results about mechanical knowing agents from mathematical logic.

show abstract

Section: An Elegant Definition Of Mathematical Knowledgementioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Short-Circuiting the Definition of Mathematical Knowledge for an Artificial General Intelligence

Alexander

2021

Software Engineering and Formal Methods. SEFM 2020 Collocated Workshops

View full text Add to dashboard Cite

show abstract

“…As Thomas Sheridan puts it 2016 [3]: "All robots for the foreseeable future will be controlled by humans", which is a strong antithesis to the "singularity" thesis [4]. There are more differentiated views in modern AI Safety research [5], particularly about what "meaningful human control" is and its implications for AI safety.…”

Section: Introductionmentioning

confidence: 99%

Using Virtual Reality for scenario-based Responsible Research and Innovation approach for Human Robot Co-production

Aschenbrenner

Tol

Rusák

et al. 2020

2020 IEEE International Conference on Artificial Intelligence and Virtual Reality (AIVR)

View full text Add to dashboard Cite

This paper proposes to use Virtual Reality scenarios to explore the reaction of stakeholders within an innovation process in the context of the introduction of robots working in close collaboration with users. The goal is to design the system upfront in such a way, that it is not perceived as a threat to the worker or his/her job. Within the responsible research and innovation approach, the introduction of new technology needs to be accompanied by a careful investigation of the thoughts and feelings of all stakeholders. Especially workers who are currently not working with robots but their workspace is currently undergoing an Industry 4.0 driven transformation, experience fear, that this new technology will make their jobs redundant. On the other hand, it can be observed, that successful robot interaction processes, on the one hand, increase the overall productivity, but also can enhance human well-being. The feeling of "teamwork" with the artificial intelligence entity can develop to be equally positive and motivating. To be able to design future workspaces which will result in a "teamwork" perception instead of the "fear" perception, the use of VR can be applied.

show abstract

“…Alternatively to conceiving AI safety as a twofold subject dealing with value alignment and control problem, it can be apprehended in risk-centered and cybersecurity-oriented [11]. taxonomic terms.…”

Section: Introductionmentioning

confidence: 99%

Hybrid Cognitive-Affective Strategies for AI Safety

Aliman¹

Self Cite

View full text Add to dashboard Cite

As the first author of the underlying paper, I had a vital contribution and it was solely my responsibility to write down the content and to perform an extensive literature research as well as in-depth analysis. This chapter is based on a slightly modified form of the publication: N.-M. Aliman and L. Kester. Augmented Utilitarianism for AGI Safety. In International Conference on Artificial General Intelligence, pages 11-21. Springer, 2019. As the first author of the underlying paper, I had a vital contribution and it was solely my responsibility to write down the content and to perform an extensive literature research as well as in-depth analysis.

show abstract

Error-Correction for AI Safety

Cited by 7 publications

References 23 publications

Short-Circuiting the Definition of Mathematical Knowledge for an Artificial General Intelligence

Short-Circuiting the Definition of Mathematical Knowledge for an Artificial General Intelligence

Using Virtual Reality for scenario-based Responsible Research and Innovation approach for Human Robot Co-production

Hybrid Cognitive-Affective Strategies for AI Safety

Contact Info

Product

Resources

About