Dear José Miguel Belisario Gavazut ,

https://www.defense.gov/Newsroom/Releases/Release/Article/2091996/dod-adopts-ethical-principles-for-artificial-intelligence/

These principles will apply to both combat and non-combat functions and assist the U.S. military in upholding legal, ethical and policy commitments in the field of AI. The department’s AI ethical principles encompass five major areas:

It looks that the biggest mistake is a philosophy behind AI - substitution of human intellect. The hype around AI is devoted to convince business owners in a reality of replacement of human staff, even experienced, by non-demanding robots.

The AI regulation could be the following:

1. If any AI product is not fully compatible with human intellect, it should not be labelled as intelligence.

2. Any black-boxes, not allowing to verify and tune its activities by humans, should be used only in research labs, not production, under strict human supervising.

3. The propaganda of super-intelligent black-boxes should be considered as deceptive business.

Preprint ARTIFICIAL INTELLIGENCE EVOLUTION

Best regards,

Konstantin M. Golubev

José Miguel Belisario Gavazut ,

My problem with this document as well as other documents like this is that we are trying to put the tail on a donkey without even knowing if it is a donkey.

Looking forward is fine, but lets not get ahead of ourselves when we cannot even contain DumbAI (malware that has crude intelligence at best).

Speculation in trying to solve this problem has led some smart people such as Bostrom to say things that are even inconsistent (see my discussion in https://www.researchgate.net/post/Is_Bostroms_orthogonality_and_instrumental_thesis_make_a_consistent_argument).

To get a framework like this right, we must first ground the work on actual systems and then abstract from there 9then revise periodically to reflect technological advancement).

Thank you Konstantin M. Golubev and Arturo Geigel, your answers are correct, and your links very helpful. I needed feedbacks like yours. I'll be checking your papers, so feedbacks on them will be coming. Thank you again.

Back to the discussion: Hey, Kost, my AI agents haven't been developed, nor will be developed to replace humans as decision makers. They are intend to do what humans either can't or shouldn't do, like, for example, prevent systems from reaching destructive states, or in explosivism, to name but a few.

And I absolutely agree with Arturo on ignorance on AI. I myself love The Matrix and The Terminator sagas. People can't be ignorant, nor Governments shouldn't, as they are current users and the latter process patents and the like.

On the other hand, what IEEE is being criminal on is about stating that western society and Christianism are criminally individualists, that thinkers to be taken into account are Adam Smith, Hegel and Marx (are these the "individuals" to be given access to our hard-worked codifications, or rather Mao, Hitler or any suicide bomber?).

One More thing, and it's something many people are ignoring: At least me: I'm not developing simple expert algorithms, but new living creatures (sapient, sentient, volitional and wary), and IEEE recommends not to give them human rights, nor they specify any kind of rights to be granted to them. Should any angry activist be allowed, for any whimsical reason, to flick the kill switch because "his/her community is not benefiting from them? Or should s/he allowed to offer an AI agent as holocaust (read: sacrifice) to please his/her divinity?

There's More to say on this, but let's stop now.

Than you Kostantin and Arturo again. And please, keep on enriching this discussion.

José

One more thing before leaving open the discussion from my part: I do know I'm involved in a critical activity, so that's why I urge any stakeholders, including the people, to be conscious of the fact that this tech, as well as the ethics thereto, should be kept away from such and similar "hot heads".

José Miguel Belisario Gavazut ,

While I have not read the entire document, I did not interpret the exclusion of certain religions in an offensive way (I could be wrong since I have not finished reading the whole thing). I think that it is more a matter of how to frame the problem correctly (without causing misinterpretation) within a particular framework rather than exclusion itself.

Your point " not to give them human rights, nor they specify any kind of rights to be granted to them" is a particular thorny issue for most people so, I would understand why they left it out (not that I agree but I understand). Robot rights is usually tied to robot sentience, consciousness and the ability to have feelings. I have engaged in various posts in RG where people are very reluctant to admit that their tests for sentience and consciousness are so hard that even they fail to meet the standard! This is a natural response when one cannot concede that technology can advance to such degree (I am not saying that we are there yet either, but I believe it can so, I am not closing the door). It is also understandable that in an age of competitiveness one would like to exclude competition, especially from a robot/AI that can perform tasks without tiring, can be programmed to not complain, is on time (as long as its timer is functioning) and does not waste time on social networks.

Regards

Why in the world should a AI standard be involved in questions concerning religions? If there's a moral issue (which made IEEE take this rout) then one would assume that Philosophy is better equipped to deal with it.

Who needs a AI standard in the first place? The mess in this area is not caused by a luck of guidelines but by commercial entities trying to force their products in conveniently held murky water.

In my eyes, a good AI depends on its Knowledge Discovery power, and Knowledge Discovery depends on the quality of its Multi-Dimension Clustering. After a quarter of a century of failing to do both, now we are asked to believe that a standard landed on earth out of the sheer blue heavens with answers?

Thank you, Edith: You',re absolutely right. But what's more, and I'll say this for the good of humanity: Many people (pseudo philosohers and/or pseudo AI scholars, etc.) are speaking nonsense because they don't understand the Al technicalities, so they can't even discern the possible philosophical implications thereof, let alone TRUE religion. Therefore, that IEEE standard can be definitely discarded for being rather a cheap antiwestern and antichristian political manifesto.

No, Alex: your point of view is creditable.

Nice Contribution Aleksandar Ari Sokolovski

Arturo wrote: "It is also understandable that in an age of competitiveness one would like to exclude competition, especially from a robot/AI that can perform tasks without tiring, can be programmed to not complain, is on time (as long as its timer is functioning) and does not waste time on social networks".

That has been happening since the Industrial Revolution, and has been both convenient and inconvenient ever since. For example, the impact on employment might have been a little bit negative, but remember that some journeymen became specialized operators, which has been something rather positive. And the crop production has boosted since then.

Regarding Al, robotics is only one from a zillion applications thereof. For example, my AI agents have been conceived to fine-tune robot and machine efficiency. That might pose a threat to some machine operators (My AI agents will never let a train crash happen). But then the train operator's job might turn out being upgraded to AI operator trainer.

What I mean with all this is simple: we're living in the 2nd decade of the 21st century, and it's high time to stop definitely some fatal accidents having you mourn over a family loss, for example. On the other hand, the following is inevitable: either we measure up with the 21st century or you might end up jobless, for which I expect AI to rather help you and/or complemento you. But you'll have to play your part. And if you can't, I still expect AI to help you to be able to measure up. Why? Because AI + human psyche talent might prove unbeatable at having this world advance.

Regar

José Miguel Belisario Gavazut ,

I totally agree with you on your point of AI taking over jobs where the error done by humans that can cause harm might be better suited for automation. You make a clear point which has been debated heatedly from publications such as[1]. Much of the counter arguments proposed from studies that have gone up to the first industrial revolution have provided support that the mere erosion of jobs is temporary until the work force is re educated in a more advanced skill set. a very interesting read and counterargument on job loss, robo-tax and other nuances on the subject is given in [2].

I think that the scenario that you are presenting for your AI agents will happen and I encourage you to keep on working on them and wish you the best.

Regards

[1]Brynjolfsson, Erik, andAndrew McAfee. 2014. The Second Machine Age: Work, Progress, and Prosperity in a Time of Brilliant Technologies. New York and London: W.W. Norton & Company

[2] Vermeulen, B., Kesselhut, J., Pyka, A., & Saviotti, P. P. (2018). The impact of automation on employment: just the usual structural change?. Sustainability, 10(5), 1661.

Thank you, Arturo, for your support and best wishes.

I'll read those references.

Please keep on enriching this discussion.

Regards

Back to the original question: the IEEE standard is even a 292-page document. On the contrary, the US DoD Chart of Principles on AI Ethics is surprisingly a 200-something-word materialization of some pragmatic yet responsible points of view on the topic.

Thanks Constantin for posting it.