Reasons and Essays

Real Time Delphi

Collecting and Sythesizing Expert Opinioni

Please provide the rationale for your answers on the previous questionnaire page.

To submit your response

To return to the questionnaire

Row	Question	Your Rationale	Group Responses
1	Here are some factors that might be included in the design of a UN AGI governance agency. Please rate them from 1 to 10, where: 10= Essential, must include 5= Possibly include 1= Destructive, must exclude Please share some of your reasoning behind your judgments at the bottom of each column. .....		*() The flight recorder is a good idea. Before the concept was used in airplanes, the cause of airplane crashes was much harder to determine. For AGI's it will be useful in identifying intrusions and reasons for operatung outside of approved limits, () Standards for certification are a basic, essential aspect of deploying any new capability... or should be. I give high marks to safety and alignment to international values, standards, etc. However, I give just average scoring to continuous audits and alignment to societal values and priorities. It is not that they are not good objectives, they are, but continuous audits by whom? What organization is a truly impartial judge? Similarly, what user is going to keep a log of AGI use, and what organization receives all this data? () Continuous audit be imbedded software. I worry that some may rate an item - should vs. how likely. I might rate something a 10 that it should be part of the UN's certification of a nation's licensing system, but rate it 5 on likelihood. () Today anyone with an advanced gaming computer can develop and deploy AI systems, and as computers continue to advance in speed and memory size, as long as people are allowed to own computers they will be able to implement and deploy AGI systems without worrying about certification or even disclosing that the systems exist. Bad actors will act badly. () It is difficult to see what would constitute "proof" of alignment. We do not even have a mechanism to ascertain this in human-to-human interactions, so what would it be in the case of human-to-AGI? It is also important to realize that we already live in a world full to the brim with NGIs (natural general intelligences) in the form of our animal relatives. To discuss the governance of AGI is to question our tacit agreements about how we treat other already-existent beings whose consciousness is unlike ours. Perhaps the parallel here is between natural intelligence and ANI; regardless, basing the distinction of which we are discussing on the simple basis that one is "artificial" while the other is not is... artificial. For these and other reasons I speak of elsewhere, the problem being discussed here is not actually about "governance of AGI" but simply about "governance." () The role of the AGI safety auditor is key. () requiring users to keep a log is an open door for privacy violations. () To me, these all seem very important. That said, I worry that one result is high bureaucratisation. That can too easily lead to inflexibility, and low responsiveness. I’m unsure that can be done about that. As far as I can tell, so far there does not seem to be any encouragement or mechanisms for bottom-up engagement also to exist. I’m unsure what can be done. And I don’t know if the sections I haven't yet looked at cover this issue. () Deep fakes and disinformation will be only a first issue to be addressed and prevented. More complex ones will arise with AGI. () a system that does not stifle innovation, creativity and development, but is monitored and supervised, but not so much that it stifles it () Certification is an important matter for the assessment of AGI systems. However such a complex matter requires specicific reasearch and investigation. () Proving safety as part of the initial audit is essential but "safety" is too vague at present. Initially I suggest defining "red lines" for obviously unacceptable behaviours such as self-replication and advising terrorists on bioweapon design, and requiring formal proofs of compliance. () A lot of the points here assume that decision-making authority has been given to a system, and that the system is responsible - and not those that made it and those that use it. Without human liability for what humans make and humans use, there is NO governance of technology. () I gave only the 5s because I do not refer to the institutional issues of governance for two reasons: First, I do not feel competent. Second, the problems are more profound. An example: If any actors utter: “It’s fake” it can be assumed that it is also a second-level fake. It brings about the question of controlling those who declare that something is fake. It is only briefly mentioned, but it is becoming even more difficult with the deep fakes. () There is no way to enforce any licensing when it will soon be possible to implement AI on systems that can be bought as child's toys. Pretending otherwise is a waste of money. Some specific licensing (ie, to access an AI via a website or corporate portal) may be valid. There might be reasonable licensing schemes for specific AI's (as opposed to AGI) where problems of alignment and safety MAY be tractable. In the general case we have no knowledge of how to implement either, how to legislate them, or how to determine compliance. Internal flight-recorder style logs are a good idea, but again, enforcement would be next to impossible, unless the AI is deployed in some public arena. Seeing as how very little before-hand prevention will be possible, I do believe that laws establishing liability and accountability will need to exist and be enforced. However, I fear that they will be hamstrung by the legal/political establishment. After all, should not a governing body be held accountable for making laws that cause more harm than good? With respect to controversial technologies, this is almost always the case, due to the lack of (and lack of respect for) knowledge of the technologies being legislated. () Stringent Security and creation and use of deepfakes are more so functions of cyber infrastructure and the field of AI as a whole and not specific to AGI and could not be considered in this form. Proof of alignment is quite difficult to guarantee and quantify. International agencies usually derive their values and functions based on collaboration, relationships and sovereign equality, how would an AGI suppose alignment with its functions based on these principles. This is quite difficult to realize, instead principles of safety, inclusion and autonomy should be aligned. () "Proof of alignment with agreed international principles, values, standards such as those of UNECSCO, OECD, Global Partnership on AI, ISO, and IEEE.". Current standard and other bodies listed here are so slow that it takes years to approve a standard and foreman opinion. Such a requirement is not possible to be implemented not enforced. () Aunque la AGI es un fenómeno de alcance global y supranacional, me parece muy difícil que la ONU pueda ejercer un control efectivo sobre lo que ocurre en Estados soberanos. Más aún, aunque los Estados lo quisieran, en la práctica es imposible; de hecho, hay medidas de control en otros campos más simples y fáciles de ejercer, que la ONU no ha logrado implementar con eficiencia. En consecuencia, cualquier medida de control o coercitiva que se pretenda incluir en una agencia de gobernanza de la ONU, será más bien declarativa y quedará sujeta a la buena voluntad de los países y, en ningún caso, tendrá efecto alguno en otros actores, especialmente si ejecutan desarrollos con fines poco éticos. Sugiero que el tono esté más en la convicción, la persuasión, la concientización y el desarrollo de una cultura del buen uso de la AGI, que en lo coercitivo. () The intention behind item 1.2.4 probably corresponds with requiring ISO/IEC 27001 and other corporate security standards certifications. Due to the high cost and efforts required for such certifications, probably a lighter mechanism (e.g. labeling) could be implemented. Since there are industries where deep fakes could be useful, e.g. in cinema, books publishing (genres like fantasy, sci-fi and history), education, etc., item 1.2.5 may not provide any gain. () It cannot be regulated excessively, it can inhibit the development of the application of artificial intelligence for the benefit of people. () Mostly the preparation of a secure civil society UN AI node will depend upon the active participation of military agencies. Nations will not contribute to the UN effort unless national security is also addressed within the context of equitable global security. A parallel AI military program should be put in place through the UN security council. Global cyberspace needs to be secure, this is a critical and pressing issue in particular because upcoming levels of quantum computing will make current techniques of encryption obsolete. Thereby rendering national grids, banking and essential services entirely vulnerable. Consensual security systems need to be developed. These can be generated at Treaty level, open ended and optional, but dedicated in spirit to the reduction of international tensions, process against untoward technological escalation and a further arms race and effective mobilization against crisis or terrorism. If the UN security council does not address AI and the cyberspace paradigm, this arena is likely to become a detrimental factor in international relations, trending to the segregation of national interests. *() I believe we can take the lessons learned about human subject protections, resulting in the creation of IRBs and apply it to AGIs. But we need a more wholistic perspective. I typed in a model earlier, but I'm not sure the information was retained.

To submit your response

Please close this window to return to your place in the questionnaire.

To sign out click here

Date: 2024-05-08