Talk:Regulation of algorithms

Recommend deletion

Geysirhead The only WP:RS I can find on "regulation of algorithms" as an umbrella term, are the [1] link already in the document and "An FDA for algorithms" 'Tutt, Andrew. "An FDA for algorithms." Admin. L. Rev. 69 (2017): 83.'. IMHO recommend deletion unless and until the term "regulation of algorithms" gains greater usage. It would be better to focus effort on improving the existing pages on related topics (again IMHO). Rolf H Nelson (talk) 18:04, 22 March 2020 (UTC)

Thank you for feedback! This is indeed an umbrella term, which is mentioned in two separate contexts on Wikipedia: "right for explanation" and "existential risk of AI". It should not be confused with "algorithmic regulation" or "regulation by algorithms", which is quite the opposite. I suppose that the creation of laws and rules about algorithms do not only appear under this term. If I delete this article, I will need to interconnect "right for explanation" with "existential risk of AI" and distinguish it from "algorithmic regulation" — this is also a suboptimal solution. Concrete suggestions are welcome! Geysirhead (talk) 18:22, 22 March 2020 (UTC)

OK, I didn't get the disambiguation. I've changed my mind, a page about the "regulation of algorithms" (whatever it's titled) has enough sources to pass WP:NOTABILITY (based on searching instead for "algorithmic regulation") for a small or mid-sized umbrella or parent article. Rolf H Nelson (talk) 21:20, 22 March 2020 (UTC)

I have undone the merge of Regulation of AI and Regulation of algorithms; no consensus was established. Geysirhead, you're going to have to justify it here. One obvious problem is that 'Regulation of artificial intelligence" is the more common term according to Google. It's also common in the literature, for example see this: https://www.cambridge.org/core/services/aop-cambridge-core/content/view/AF1AD1940B70DB88D2B24202EE933F1B/S1867299X19000084a.pdf/towards_intelligent_regulation_of_artificial_intelligence.pdf. Compare "regulation of artificial intelligence" and "regulation of algorithms" on Google Scholar. I am not saying the pages could not be merged, only that we need consensus. Johncdraper (talk) 10:36, 28 March 2020 (UTC)

There is no clear definition of the term intelligence. Therefore, we have the unclear term Artificial Intelligence, where only the term Artificial is clear and the rest is changing over time. Is a search engine using Tf–idf artificial intelligence? Now, it is not. But, it is was once. The term algorithm is much more clear. Most people understand Strong AI or at least artificial general intelligence if they hear about AI. There are also too many stories and movies, which confuse AI with Artificial Life. Most fictional AI systems mimic mammals if they follow their goal of self-preservation — instead of self-copying in networks, fictional AIs grasp for resources and eliminate human opponents. Nowadays, "Artificial Intelligence" is also a confusing marketing term for artificial neural networks or intelligent agents (“The key difference, is that an algorithm defines the process through which a decision is made, and AI uses training data to make such a decision." CEO of QuiGig). In Summary, AI might be (mis)understood as:

• Historically simple but useful algorithms like Tf–idf
• Strong AI
• Artificial general intelligence
• Artificial Life based on computer hardware mimicking mammals
• Artificial neural networks
• Intelligent agent

IMO, it is impossible to draw a line between "intelligent algorithms" and "non-intelligent algorithms". There is no clear example of a "non-intelligent algorithm" if such a thing actually exists. Otherwise, if one says Regulation of algorithms, it is clear to everybody, what is meant. Regulation of artificial intelligence could sound in some ears like "suing robots". I agree that sources use terms in confusing ways. All I want is clear this confusion.Geysirhead (talk) 11:28, 28 March 2020 (UTC)

Geysirhead Okay, so I think we can agree they are very related concepts, and from a computing perspective, I agree with you. No algorithm is truly 'intelligent' until we get to conscious, elf-aware, ethical AGI agents. There is, to the purist, the whole problem of whether AI is an ideology. In any case, logically, 'Regulation of AI' is a subset of 'Regulation of Algorithms', so I understand why you merged. However, in developing a page on 'Regulation of Artificial Intelligence', I was influenced by general usage: a) There is a Wikipedia page on both 'Artificial Intelligence' and 'Algorithm', so Wikipedia practice is to distinguish, and b) emerging practice, e.g., by the Library of Congress Law Library, see here: https://www.loc.gov/law/help/artificial-intelligence/regulation-artificial-intelligence.pdf. This reflects the fact that governments are developing the term 'Regulation of Artificial Intelligence', e.g., here: https://www.federalregister.gov/documents/2020/01/13/2020-00261/request-for-comments-on-a-draft-memorandum-to-the-heads-of-executive-departments-and-agencies. This is based on dictionary definitions employed in such documents, so e.g., the Library of Congress uses the Oxford Dictionary “the theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages.” One way forwards may be to keep both pages but to properly cross-reference. You make 'Regulation of Artificial Intelligence' a subentry under 'Regulation of Algorithms'. Then, on the 'Regulation of Artificial Intelligence' page, I clarify that it is a subset of 'Regulation of Algorithms'. What do you think? Johncdraper (talk) 11:59, 28 March 2020 (UTC)

Well, one can either agree that no algorithm is truly 'intelligent' or that all algorithms are intelligent. But, if we are talking about "human intelligence" as in your citation, one can determine the "AI algorithms" subset inside of general algorithms. Yes, I agree with your suggestion.Geysirhead (talk) 12:25, 28 March 2020 (UTC)

Johncdraper I created the subsection and linked your article right in the beginning. And, "regulation of blockchain algorithms" is the second subset of "regulation of algorithms".Geysirhead (talk) 13:37, 28 March 2020 (UTC)

Geysirhead Okay, I will work on referring to 'Regulation of Algorithms' in the 'Regulation of Artificial Intelligence' page. On 'Regulation of Algorithms', see also regulation of high frequency trading algorithms: https://en.wikipedia.org/wiki/High-frequency_trading#Regulation_and_enforcement. That seems relevant to the 'Regulation of Algorithms' as it concerns regulations of both humans and the algorithms employed.Johncdraper (talk) 14:04, 28 March 2020 (UTC)

Thanks! HFT shifts into AI domain. :-) Geysirhead (talk) 14:52, 28 March 2020 (UTC)

Given we have agreement on the relationship between the two pages, I propose removing the 'merge' template from the 'Regulation of Artificial Intelligence' page. Johncdraper (talk) 14:26, 29 March 2020 (UTC)

... So, delete or not delete?? — Preceding unsigned comment added by 201.81.201.3 (talk • contribs) Not delete. Rolf H Nelson (talk) 03:05, 15 June 2020 (UTC)

Regulation to mitigate the harms and biases of AI

The Algorithmic Justice League, proposes a regulation of algorithms, to mitigate the harms and biases of AI, by promoting 4 core principles.

"

• Affirmative consent: everyone should have a real choice in how and whether they interact with AI systems.

• Meaningful transparency: it is of vital public interest that people are able to understand the processes of creating and deploying AI in a meaningful way, and that we have full understanding of what AI can and cannot do.

• Continuous oversight and accountability: politicians and policymakers need to create robust mechanisms that protect people from the harms of AI and related systems both by continuously monitoring and limiting the worst abuses and holding companies and other institutions accountable when harms occur. Everyone, especially those who are most impacted, must have access to redress from AI harms. Moreover, institutions and decision makers that utilize AI technologies must be subject to accountability that goes beyond self-regulation.

• Actionable critique: we aim to end harmful practices in AI, rather than name and shame. We do this by conducting research and translating what we’ve learned into principles, best practices and recommendations that we use as the basis for our advocacy, education and awareness-building efforts. We are focused on shifting industry practices among those creating and commercializing today’s systems.

" — Preceding unsigned comment added by 201.81.201.3 (talk • contribs)

Algorithmification of regulations

The Regulation of AI is not a synonym for "regulation of algorithms", because AI algorithms are only a small subset of all algorithms. In general the AI algorithm is used to implement an intelligent agent to replace a human in a subjective judgement.

By other hand "non intelligent" algorithms can be viewed as a formal step-by-step transformation of law into a software. The first step is to translate "free text" law into something like Legal XML or HTML+RDFa markup... So, in this context "Regulation of algorithms" is a set of standards (including text formats/markups), and rules about when is valid to transform some formalized law in to software, when not.

— Preceding unsigned comment added by 201.81.201.3 (talk • contribs)

The article doesn’t say that regulation of AI is a synonym for regulation of algorithms. It does say that when the specific subset of AI algorithms are in question, the terms commonly used is “Regulation of AI”. I thing the current text is clear enough. The rest of the request is a non sequitur. Ferkijel (talk) 12:10, 28 March 2021 (UTC)

Some Proposed Changes

Hello, I am employed by Boston University's Fineman & Pappas Law Libraries. After reviewing this Wikipedia page, I believe that information from one of our faculty's scholarship might provide a valuable addition to this page. I would appreciate it if this requested edit could be reviewed.

Implementation

Add to the end of the first paragraph under the implementation section:

There are many concerns that there is not enough visibility and monitoring of AI in these sectors. In the financial sector, for example, there have been calls for the Consumer Financial Protection Bureau to more closely examine source code and algorithms when conducting audits of financial institutions' non-public data.— Preceding unsigned comment added by Cf2022 (talk • contribs) 18:20, December 21, 2020 (UTC)

References

Cf2022 (talk) 04:13, 6 January 2021 (UTC)Cf2022

 Done Ferkijel (talk) 10:59, 28 March 2021 (UTC)

Mention of "The Social Dilemma" ?

https://en.wikipedia.org/wiki/The_Social_Dilemma — Preceding unsigned comment added by 2600:8804:6600:592:516E:FC4F:6E52:B4F7 (talk) 21:31, 7 January 2021 (UTC)