Jump to content

Talk:Machine ethics

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

Wiki Education Foundation-supported course assignment

[edit]

This article was the subject of a Wiki Education Foundation-supported course assignment, between 27 August 2019 and 6 December 2019. Further details are available on the course page. Student editor(s): Goflores, Avnishna, Najimene. Peer reviewers: Hannibalrising94, Ahnmelis, BorizAlva.

Above undated message substituted from Template:Dashboard.wikiedu.org assignment by PrimeBOT (talk) 00:23, 18 January 2022 (UTC)[reply]

Partisan Sources and Tone

[edit]

The article is in dire need of copyediting on the tone, style, and balance of POV. There is significant research in this field well beyond the cluster of sources that are being presented here. I plan to assist in development of this page based on solid, independent sources. As it stands, several sections, if not the entire article, need to be rewritten completely and balanced out. It appears to present two or three authors, who have a working relation (thereby making it a non-independent set of sources as per WP:IS) and are presenting their views with WP:UNDUE weight. --Lightbound talk 19:25, 31 March 2014 (UTC)[reply]

What do you have in mind by "research in this field"? Armchair philosophy? I'd love to see what you have to add, because it looks to me like the field is very new, and very little noteworthy has been done except to review/categorize the armchair philosophy, build machines (of questionable morality), and promote the idea of doing research. Who has advanced any plausible paradigms for how research could be conducted in this field? — Preceding unsigned comment added by 165.189.37.11 (talk) 15:29, 10 November 2014 (UTC)[reply]
@Lightbound: This article contains a large number of sources. Which individual sources have been identified as "partisan"? Jarble (talk) 22:10, 10 November 2014 (UTC)[reply]
@Lightbound: I am intending to begin a fairly large overhaul of this page and it's content and layout. There aren't a whole lot of people working in this field but there are some and I hope they can be added to this page in a meaningful way. DannCrow (talk) 18:23, 3 December 2014 (UTC)[reply]
@DannCrow: I added a section, "Artificial systems and moral responsibility" on the page for Moral responsibility but perhaps I acted too soon? Or maybe this can be moved across in due course? Spectrum629 (talk) 06:12, 15 January 2015 (UTC)[reply]
@Jarbie: Here's my take on it: Moor's 2006 article [1] is seminal and appears to coin the term "machine ethics" (thus satisfying WP:WEIGHT), the questions that Moor posed in his article are ancient (opening up to charges of WP:NPOV). Spectrum629 (talk) 07:42, 15 January 2015 (UTC)[reply]

Content added today

[edit]

Following content was added today. To me this is WP:OR. please discuss, thanks

Reasons for relevance

[edit]

Intelligence explosion

[edit]

Some researchers have argued that an intelligence explosion could take place in the near future. [2] [3] [4] The resulting AI could have very high impact, [5][6] but must not necessarily share our values.[7] It is therefore important to know how to avoid catastrophical impact from an AI that conflicts with our values before an intelligence explosion. Furthermore, a "friendly" artificial intelligence could also have an ethically positive impact. [8] [9]

Current developments

[edit]

Artificial systems have been pointed out to make ethically critical decisions already.[10]

Machine ethics as a tool for the general field of ethics

[edit]

Machine ethics and formalization of moral intuitions has been argued to inspire ethical discussion in general.[8] [11]

Criticism and counter-arguments

[edit]

Machine ethics is often criticized as not being useful right now. For example, Andrew Ng argued that accidentally building human-level "evil" AI is possible, but "just so far away that I don’t know how to productively work on that." [12]

In a review of Nick Bostrom's book Superintelligence, Ernest Davis, computer science professor at the university of New York, pointed out that if a superintelligence can understand the mind of humans, the value loading problem may become trivial, since one could simply tell the AI to do what humans want. [13]

references

[edit]

References

  1. ^ Moor, James H. (2006). "The Nature, Importance and Difficulty of Machine Ethics". IEEE Intelligent Systems. 21 (4): 18-21. doi:10.1109/MIS.2006.80. {{cite journal}}: Unknown parameter |month= ignored (help)
  2. ^ Schmidhuber, Jürgen (2012), New Millennium AI and the Convergence of History: Update of 2012 (PDF), retrieved 6 March 2015 {{citation}}: Unknown parameter |agency= ignored (help)
  3. ^ Kurzweil, Ray (2005). The Singularity Is Near. When Humans Transcend Biology. New York: Viking Books. {{cite book}}: |access-date= requires |url= (help)
  4. ^ Vinge, Vernor (1993), The Coming Technological Singularity (PDF), retrieved 6 March 2015
  5. ^ Hawking, Stephen; Tegmark, Max; Russel, Stuart; Wilczek, Frank (19 April 2014), Transcending Complacency on Superintelligent Machines, Huffington Post {{citation}}: |access-date= requires |url= (help)
  6. ^ Yudkowsky, Eliezer, Artificial Intelligence as a Positive and Negative Factor in Global Risk (PDF), New York: Oxford University Press, retrieved 6 March 2015 {{citation}}: line feed character in |title= at position 42 (help)
  7. ^ Bostrom, Nick (2012), The Superintelligent Will: Motivation and Instrumental Rationality in Advanced Artificial Agents (PDF), retrieved 6 March 2015 {{citation}}: Unknown parameter |agency= ignored (help)
  8. ^ a b Hall, J. Storrs (2011), "Ethics for Self-Improving Machines", in Anderson, Michael; Anderson, Susan Leigh Anderson (eds.), Machine Ethics, Cambridge Unviersity Press, pp. 512–523, ISBN 978-0-521-11235-2 Cite error: The named reference "Gips2011" was defined multiple times with different content (see the help page).
  9. ^ Muehlhauser, Luke, "Engineering Utopia", Facing the Intelligence Explosion
  10. ^ Wallach, Wendell; Allen (2009), Moral Machines: Teaching Robots Right from Wrong, Oxford: Oxford University Press {{citation}}: Unknown parameter |fist2= ignored (help)
  11. ^ Anderson, Susan Leigh (2011), "How Machines Might Help Us Achieve Break- throughs in Ethical Theory and Inspire Us to Behave Better", in Anderson, Michael; Anderson, Susan Leigh Anderson (eds.), Machine Ethics, Cambridge Unviersity Press, pp. 524–530, ISBN 978-0-521-11235-2 {{citation}}: line feed character in |chapter= at position 42 (help)
  12. ^ Madrigal, Alexis (27 February 2015), The case against killer robots, from a guy actually working on artificial intelligence, retrieved 6 March 2015
  13. ^ Davis, Ernest (23 October 2014), Ethical Guidelines for A Superintelligence (PDF), retrieved 6 March 2015

discussion

[edit]

I did not conduct original research for this edit. Could you maybe elaborate? (Rereading, I think the following sentence is the only one that is really problematic: "It is therefore important to know how to avoid catastrophical impact from an AI that conflicts with our values before an intelligence explosion," especially since it is not followed by a note.) -Hkfscp11 — Preceding undated comment added 15:33, 6 March 2015 (UTC)[reply]

Algorithmic Fairness/Machine Learning Bias

[edit]

The main change I plan to add is a section about "algorithmic fairness," which is a new research topic in machine learning. I go through all the relative wiki and think Machine Ethics is the most suitable one. All the new citations are recent academic papers and reliable fact reports on the machine learning bias.

Here is the proposed content:


The rapid development of AI technologies makes Algorithmic Fairness an important topic in Machine ethics.[1] Algorithms increasingly affect decisions in high-stakes tasks, such as credit card applications[2] , hiring decisions[3], college admissions, and criminal sentencing[4]. In 2015, the Obama Administration's Big Data Working Group released several reports arguing “the potential of encoding discrimination in automated decisions” and calling for “equal opportunity by design” for applications such as credit scoring.

There are concerns that these technologies may introduce new biases or perpetuate existing prejudice and unfairness, either with or without intention[4]. Despite these concerns, both research and industry practice are lagging behind[5]. One common approach in the current practice is avoiding all protected attributes such as race, color, religion, gender, disability, or family status in machine learning features. However, this is problematic due to the encoding redundancy. These protected attributes can be inferred from the other features. For example, the feature combination of zip code and income [6] may connect to race demographics.

A recent study about existing Criminal Justice Systems further confirms the importance of that algorithmic fairness[4]. At various points in the criminal justice system, including decisions about bail, sentencing, or parole, an officer of the court may use quantitative risk tools to assess a defendant’s probability of recidivism based on their history and other attributes. ProPublica analyzed a commonly used statistical method for assigning risk scores in the criminal justice system - the COMPAS risk tool [7] - and argued that it was biased against African-American defendants. In that system, African-American defendants were more likely to be incorrectly labeled as higher-risk than they were, while white defendants were more likely to be incorrectly labeled as lower-risk than they were[4].

References

  1. ^ Fairness, Accountability, and Transparency in Machine Learning. http://www.fatml.org/.
  2. ^ Zafar, Muhammad Bilal; Valera, Isabel; Gomez Rodriguez, Manuel; Gummadi, Krishna P. (2017). "Fairness Beyond Disparate Treatment & Disparate Impact": 1171–1180. doi:10.1145/3038912.3052660. {{cite journal}}: Cite journal requires |journal= (help)
  3. ^ Artificial Intelligence’s White Guy Problem https://www.nytimes.com/2016/06/26/opinion/sunday/artificial-intelligences-white-guy-problem.html
  4. ^ a b c d Machine Bias: There’s software used across the country to predict future criminals. And it’s biased against blacks. https://www.propublica.org/article/machine-bias-risk-assessments-in-criminal-sentencing
  5. ^ Pedreshi, Dino; Ruggieri, Salvatore; Turini, Franco (2008). "Discrimination-aware data mining": 560. doi:10.1145/1401890.1401959. {{cite journal}}: Cite journal requires |journal= (help)
  6. ^ Identifying Significant Predictive Bias in Classifiers https://arxiv.org/pdf/1611.08292.pdf
  7. ^ Practitioners Guide to COMPAS http://www.northpointeinc.com/files/technical_documents/FieldGuide2_081412.pdf

--Shift-3 (talk) 14:28, 28 April 2017 (UTC)[reply]

first, thanks for bringing this. i'm confirming that this is important subject matter and this article should discuss this. Some issues here that I will fix.
  • First, you have embedded external links for the Obama administration stuff, and we don't do that in Wikipedia. Please make them into refs.
  • With regard to the refs, please see [[User:Jytdog/How#Formatting_sources|here] about formatting them.
  • Please capitalize in English, not German. So not "Criminal Justice Systems" but rather "criminal justice systems".
  • Nothing is "recent" in Wikipedia. Articles have no datelines. See WP:RELTIME
  • Punctuation goes before refs, not after them
  • Drilling down a bit into refs:
    • The fatml.org ref brings no value and is just used to support the WP:EDITORIALIZING first sentence.
    • The Zafar ref is brand new and has been cited zero times. Not a good source to use. (Our mission is to present the public with "accepted knowledge" and the best place to find that is very strong secondary sources - a brand new paper that has never been cited is a not a secure place to find "accepted knowledge")
    • The Zheng paper ("Identifying Significant Predictive Bias in Classifiers") is a primary source and we shouldn't use it, as is the "Discrimination-aware data mining" paper.
    • The NYT and ProPublica refs are very strong and should drive the content. Likewise the review from the Justice department.
    • it isn't clear why Northpointe's sales document is here...
anyway I will fix this stuff. Thanks again Jytdog (talk) 17:25, 28 April 2017 (UTC)[reply]
I added bits of this to the recividism article and to the quantitative risk assessment software article too. thanks again Jytdog (talk) 21:29, 28 April 2017 (UTC)[reply]
[edit]

Hello fellow Wikipedians,

I have just modified 5 external links on Machine ethics. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:

When you have finished reviewing my changes, you may follow the instructions on the template below to fix any issues with the URLs.

This message was posted before February 2018. After February 2018, "External links modified" talk page sections are no longer generated or monitored by InternetArchiveBot. No special action is required regarding these talk page notices, other than regular verification using the archive tool instructions below. Editors have permission to delete these "External links modified" talk page sections if they want to de-clutter talk pages, but see the RfC before doing mass systematic removals. This message is updated dynamically through the template {{source check}} (last update: 5 June 2024).

  • If you have discovered URLs which were erroneously considered dead by the bot, you can report them with this tool.
  • If you found an error with any archives or the URLs themselves, you can fix them with this tool.

Cheers.—InternetArchiveBot (Report bug) 12:51, 11 January 2018 (UTC)[reply]

Difference with alignment

[edit]

I suggest to clarify somewhere the distinction between machine ethics and AI alignment. Is AI alignment a subset of machine ethics? Alenoach (talk) 19:08, 25 February 2024 (UTC)[reply]