9,542 Matching Annotations
  1. Last 7 days
    1. How to create a Topic Cluster with WordPress?

      ||Jovan|| How to create a topic cluster with WordPress?

    Created with Sketch. Visit annotations in context

    Created with Sketch. Annotators

    Created with Sketch. URL

    1. It would probably be more feasible imo to regulate the uses of a.i., for example if you wanted to utilize A.I. in some type of autonomous function, such as a self driving car, or some kind of guard robot, or traffic lights, or who knows what else, that you would need to ensure some level of standardized security and other such potential regulations.

      regulate uses or development of AI.

    2. OpenAI just keeps saying it's coming, it's dangerous, but when congress asked for "nutrition label" of their AI model, they avoided the question.

      ||Jovan|| good point

    3. To protect their positions, monopolists can:A) Influence public opinion by drawing on science fiction imagery.B) Position themselves as the first supporters of the need for regulation;C) Induce legislators to enact very restrictive regulations with them as the main interlocutors.Expected outcome:- Stringent regulations;- Strong limitations on AI open source;- Long live the new monopolists!

      ||Jovan|| This is a good summary.

    4. Actually, we would be safer with a multilarity than a singularity because others very advanced AI or parts of AI would be able to counter one very advanced AI outside the realm of humanity basic needs.

      No singularity but multilarity

    5. Open-source may post a challenge as well for global cooperation. If everyone can cook AI models in their basements, how can AI truly be aligned to safe objectives?

      ||sorina|| it is concentrated effort to stop bottom-up AI. It is very dangerous development.

    6. About 58% of U.S. adults are familiar with ChatGPT.

      ||JovanNj|| ||sorina|| ||anjadjATdiplomacy.edu|| Relatively low awareness about ChatGPT and low use (only 14%). It is interesting that Asian minorities are more activite in using ChatGPT.

    Created with Sketch. Visit annotations in context

    Created with Sketch. Annotators

    Created with Sketch. URL

    1. ||Jovan|| An important letter from European open-source community.

    Created with Sketch. Visit annotations in context

    Created with Sketch. Annotators

    Created with Sketch. URL

    1. AGI is the intelligence of a machine that can understand, learn, plan, and execute any intellectual task that a human being can.

      Q: What is Artificial General Intelligence (AGI)?

    2. In the field of AI, there's often a trade-off between safety and performance, known as the alignment tax.

      Q: what is the alignment tax?

    3. Outcome supervision involves giving feedback based on the final result, whereas process supervision provides feedback for each individual step in a process.

      ||Jovan|| Is this real brekathrough or part of current spin of OpenAI?

    Created with Sketch. Visit annotations in context

    Created with Sketch. Annotators

    Created with Sketch. URL

    1. Propaganda is picking up. You have title that does not reflect the content. In the content you have rather balanced coverage about 'AI dooms day'. It also brings a few good points such as ethical training of AI engineers.

      ||Jovan||

    2. "It's never too late to improve," says Prof Bengio of AI's current state. "It's exactly like climate change.

      Dangerous use of analogies.

    3. "We also need the people who are close to these systems to have a kind of certification... we need ethical training here. Computer scientists don't usually get that, by the way."

      Ethical training is fine. Certification, I do not know.

    4. "Governments need to track what they're doing, they need to be able to audit them, and that's just the minimum thing we do for any other sector like building aeroplanes or cars or pharmaceuticals," he said.

      Argument for centralised control.

    Created with Sketch. Visit annotations in context

    Created with Sketch. Annotators

    Created with Sketch. URL

    1. They fear that the rich world is going to tiptoe away from the SDGs, and the Summit of the Future is a sort of diplomatic smokescreen for that,”

      ||sorina||||VladaR||||Pavlina||||Katarina_An|| This article provides background about the atmosphere in New York. We should be aware of it as we prepare event in NY. We may strenghten linkages between SDGs and Future Summit.

    2. Its call for climate justice also echoes the calls by many nations, including Pakistan, to step up international funding to help countries respond to devastating climate-driven catastrophes such as the torrential storm that inundated a third of the country’s territory last year.

      climate concern of developing countries.

    3. He also foresees a potentially contentious set of negotiations over a broad range of issues, from human rights to climate justice, the environment, and a newly articulated peace agenda.

      Other issues of concern of developing countries.

    4. “There is a sense that, you know, this is an effort to change the intergovernmental structure of the United Nations and the General Assembly and the General Assembly is an organization of member states,” he said.

      Concern about developing countries on multistakeholder approach.

    5. It called for the creation of a Futures Lab to measure the impact and risks of policies over the long haul; the reform of the Trusteeship Council, established to manage decolonization, to advocate on behalf of future generations, and the appointment of a “special envoy to ensure that policy and budget decision take into account the impact of future generations.” Guterres proposed hosting a Summit of the Future this year so world leaders could turn his plan into action.

      Set of proposals for the Summit of Futre.

    Created with Sketch. Visit annotations in context

    Created with Sketch. Annotators

    Created with Sketch. URL

    1. I looked for apps containing basic editing options within the app—like highlighting, inviting teammates to comment/edit, and adjusting playback speed.

      This feature is essential for us - how to annotate transcripts.

    Created with Sketch. Visit annotations in context

    Created with Sketch. Annotators

    Created with Sketch. URL

    1. Utilizing cutting-edge fine-tuning methodologies

      ||Jovan|| test

    Created with Sketch. Visit annotations in context

    Created with Sketch. Annotators

    Created with Sketch. URL

  2. May 2023
    1. our api
    2. also think it’s really important to decide to whose values we’re going to align these models.
    3. I’m a big believer in the democratizing potential of technology,
    4. is going to be small just because of the resources required.

      Is it correct? Can we train model with less resources?

    5. I think it’s important to democratize the inputs to these systems, the values that we’re going to Alli align to.
    6. with Microsoft releasing Sydney.
    7. there are areas like copyright where we don’t really have laws.

      It is not correct. There are rules. The other question if they can be enforced.

    8. to prioritize ethics and responsible technology as opposed to posing development.
    9. My opinion is that the moratorium that we should focus on is actually deployment until we have good safety cases.
    10. try to make these things actually enforced.
    11. three principles, transparency, accountability, and limits on use.

      3 principles for AI governance

    12. These systems are almost like counterfeit people, and we don’t really honestly understand what the consequence of that is.
    13. counterfeit people
    14. Same with psychiatric advice
    15. medical misinformation
    16. haven’t been invented yet
    17. pre-deployment and post-deployment.
    18. we put so much burden that only the big players can do it.
    19. regulatory capture.
    20. you can still cause great harm with a smaller model.
    21. you slow down American industry in such a way that China or somebody else makes faster progress.
    22. economic transformation
    23. safety concerns
    24. Number one, you’re here because AI is this extraordinary new technology that everyone says can be transformative as much as the printing press. Number two is really unknown what’s gonna happen. But there’s a big fear you’ve expressed to all of you about what bad actors can do and will do if there’s no rules of the road. Number three, as a member who served in the house and now in the Senate, I’ve come to the conclusion that it’s impossible for Congress to keep up with the speed of technology.

      A good summary of the current situation with AI technology.

    25. what these systems get aligned to, whose values,
    26. There is a real risk of a kind of technocracy combined with oligarchy, where small number of companies influence people’s beliefs through the nature of these systems.
    27. this massive corporate concentration.
    28. We have API developers pay us and we have ChatGPT.
    29. we need to have some international meetings very quickly with people who have expertise in how you grow agencies in the history of growing agencies.
    30. no way to put this genie in the bottle globally.
    31. But I I, you know, you talked about defining the highest risk uses. We don’t know all of them. We really don’t. We can’t see where this is going regulating at the point of risk.
    32. we’re not an advertising based model.

      But BING is!

    33. the safety for children of you
    34. tools that humans use to make human judgments, and that we don’t take away human judgment
    35. can predict future human behavior is potentially pretty significant at the individual level.
    36. a model that could help create novel biological agents would be a great threshold.
    37. capability thresholds

      good point, but difficiult to define.

    38. Andwe risk if we’re not thoughtful about it contributing to the development of tools and approaches that only exacerbate the bias and inequities that exist in our society.

      Valid point about inequality.

    39.  Excited to work with people who have particular data sets and to work to collect a representative set of values from around the world to draw these wide bounds of what the system can do.
    40. Can can you speak just for a second specifically to language inclusivity?

      Many good questions from senators asking for clarity. There are not clear answers always. But clarity of language must prevail even if you disiplay trade-offs.

    41. language and cultural inclusivity

      Another important topic.

    42. And what auto GPT does is it allows systems to access source code, access the internet and so forth. And there are a lot of potential, let’s say cybersecurity risks. There, there should be an external agency that says, well, we need to be reassured if you’re going to release this product that there aren’t gonna be cybersecurity problems or there are ways of addressing it.

      ||VladaR|| Vlada, please follow-up on this aspect on AI and cybersecurity.

    43. the central scientific issue

      Is it 'scientific issue'? I do not think so. It is more philosophical and possible even, theological, issue. Can science tell us what is good and bad?

    44. I can’t envision or imagine right now what kind of a licensing scheme we would be able to create to pretty much regulate the vastness of, of the, this game-changing tool.

      difficult in establishing AI licencing scheme.

    45. some sort of standard,
    46. what do you consider a harmful request?

      Critical issue.

    47. require independent audits.
    48. specific tests that a model has to pass
    49. a set of safety standards
    50. a new agency that licenses any effort above a certain scale of capabilities and can take that license away and ensure compliance with safety standards.
    51. basically focus on AI safety research.
    52. ai, constitution
    53. a nimble monitoring agency to follow what’s going on.
    54. please don’t just use concepts. I’m looking for specificity.

      Great comment for AI debate

    55. So disclosure of the data that’s used to train AI, disclosure of the model and how it performs and making sure that there’s continuous governance over these models.

      Q: What are the main aspects of AI transparency?

    56. the I P C A UN body

      The closest analogy is with IPCC

    57. So we have existing regulatory authorities in place who have been clear that they have the ability to regulate in their respective domains. A lot of the issues we’re talking about today span multiple domains, elections, and the like.

      How to use existing governance and regulatory agencies?

    58. Guardrails need to be in place.

      Guardrails are increasing in 'lingo intensity'

    59. Different rules for different risks.

      Good slogan

    60. the conception of the EU AI Act is very consistent with this concept of precision regulation where you’re regulating the use of the technology in context.

      EU AI Act uses precise regulation of regulation AI in specific contexts.

    61. we need to give policy makers and the world as a whole the tools to say, here’s the values and implement them.

      Use SDGs as guardrails for AI.

    62. that interaction with the world is very important.
    63. constitutional AI

      New concept?

    64. a reasonable care standard.

      Another vague concept. What is 'reasonable'? There will be a lot of job for AI-powered lawyers.

    65. Thank you, Mr. Chairman and Senator Hawley for having this. I’m trying to find out how it is different than social media and learn from the mistakes we made with social media. The idea of not suing social media companies is to allow the internet to flourish. Because if I slander you you can sue me. If you’re a billboard company and you put up the slander, can you sue the billboard company? We said no. Basically, section 230 is being used by social media companies to high, to avoid liability for activity that other people generate. When they refuse to comply with their terms of use, a mother calls up the company and says, this app is being used to bully my child to death. You promise, in the terms of use, she would prevent bullying. And she calls three times, she gets no response, the child kills herself and they can’t sue. Do you all agree we don’t wanna do that again?

      How to avoid repeating with AI governance what happened with Seciton 230 and social media governance?

    66. And so the quality of the sort of overall news market is going to decline as we have more generated content by systems that aren’t actually reliable in the content they’re generated.

      Risks for newsmarket.

    67. the current version of GPT-4 ended to training in 2021.

      2021 starts to being 'safety net' for OpenAI

    68. And other countries are doing this, Australia and the like. And so my question is, when we already have a study by Northwestern predicting that one-third of the US newspapers are that roughly existed, two decades are gonna go, are gonna be gone by 2025, unless you start compensating for everything from book movies, books. Yes. but also news content. We’re gonna lose any realistic content producers. And so I’d like your response to that. And of course, there is an exemption for copyright in section two 30. But I think asking little newspapers to go out and sue all the time just can’t be the answer. They’re not gonna be able to keep up.

      Q: How to protect newspapers and content producers?

    69. When some of those fake ads. So that’s number one. Number two is the impact on intellectual property.

      Two concers: fake adds and IPRs.

    70. Sen. Marsha Blackburn (R-TN):

      It is probably the most practical approach to AI governance. Senator from Tennessee asked many questions on the protection of copyright of musicians. Is Nashville endangered. The more we anchor AI governance questions into practical concerns of citizens, communities, and companies - the better AI governance we will have.

    71. OpenAI Jukebox

      Diplo team shoudl follow on this development.

    72. And I wanna come to you on music and content creation, because we’ve got a lot of songwriters and artists. Yeah. And a, I think we have the best creative community on the face of the Earth. They’re in Tennessee, and they should be able to decide if their copyrighted songs and images are going to be used to train these models. And I’m concerned about OpenAI’s jukebox. It offers some renditions in the style of Garth Brooks, which suggests that OpenAI is trained on Garth Brooks songs. I went in this weekend and I said, write me a song that sounds like Garth Brooks. And it gave me a different version of Simple Man. So it’s interesting that it would do that. But you’re training it on these copyrighted songs, these mini files, these sound technologies. So as you do this, who owns the rights to that AI generated material and using your technology, could I remake a song, insert content from my favorite artist, and then own the creative right to that song?

      Bring intellectual property into debate

    73. that people own their virtual you.

      People can own it only with 'bottom-up AI'

    74. We’ve done it before with the IAEA.

      Now IAEA comes as analogy, probably driven by nuclear power?

    75. When you think about the energy costs alone, just for training these systems, it would not be a good model if every country has its own policies and each, for each jurisdiction, every company has to train another model.

      It is naive view because AI is shaped by ethics and ethics is very 'local'. Yes, there are some global ethical principles: protect human life and dignity. But many other ethical rules are very 'local'.

    76. need a cabinet level organization within the United States in order to address this.

      Who can govern AI?

    77. So we think that AI should be regulated at the point of risk, essentially, and that’s the point at which technology meets society.

      Nice 'meeting' language

    78. I can’t recall when we’ve had people representing large corporations or private sector entities come before us and plead with us to regulate them. In fact, many people in the Senate have base their careers on the opposite that the economy will thrive if government gets the hell out of the way. And what I’m hearing instead today is that ‘stop me before I innovate again’ message. And I’m just curious as to how we’re going to achieve this

      Great point and strategic shift. It is 'Frankenstein moment'. Companies realised that they created something they cannot control.

    79. an enterprise technology company, not consumer focused. S

      It is an interesting distinction. However, technology developed by IBM will be used for consumer services.

    80. hyper targeting of advertising is definitely going to come.
    81. And we probably need scientists in there doing analysis in order to understand what the political influences of, for example, of these systems might be.

      Markus tries to make case for 'scientists'. But, frankly speaking, how scientists can decide if AI should rely on book written in favour of republicans or democrats or, even more as AI develops with more sophistication, what 'weight' they should give to one or another source.

      It is VERY dangerous to place ethical and political decisions in hands of scientists. It is also unfair towards them.

    82. we don’t know what it’s trained on.

      It is a good point. OpenAI is not transparent on datasets used for training GPT. But, the problem is that even if they inform us, the question will be who decided what datasets should be used for training.

    83. If these large language models can, even now, based on the information we put into them quite accurately predict public opinion, you know, ahead of time. I mean, predict, it’s before you even ask the public these questions, what will happen when entities, whether it’s corporate entities or whether it’s governmental entities, or whether it’s campaigns or whether it’s foreign actors, take this survey information, these predictions about public opinion and then fine tune strategies to elicit certain responses, certain behavioral responses.

      this is what worries politicians - how to win elections? They like 'to see' (use AI for their needs) but 'not to be seen' (use by somebody else. The main problem with political elites worldwide is that they may win elections with use of AI (or not), but the humanity is sliding into 'knowledge slavery' by AI.

    84. large language models can indeed predict public opinion.

      They can as they, for example, predict continuation of this debate in the political space.

    85. so-called artificial general intelligence really will replace a large fraction of human jobs.

      It is a good point. There won't be more work.

    86. And the real question is over what time scale? Is it gonna be 10 years? Is it gonna be a hundred years?

      It is a crucial question. One generation will be 'thrown under the bus' in transition. Generation of age 25-50 should 'fasten seat-belts'. They were educated in the 'old system' while they have to work in a very uncertain new economy.

    87. that scientists be part of that process.

      What should scientist do specifically? Can scientist judge if something is true? Who are scientists (e.g. do we refer to IT specialists)?

    88. So I think the most important thing that we could be doing and can, and should be doing now, is to prepare the workforce of today and the workforce of tomorrow for partnering with AI technologies and using them. And we’ve been very involved for, for years now in doing that in focusing on skills-based hiring in educating for the skills of the future. Our skills build platform has 7 million learners and over a thousand courses worldwide focused on skills. And we’ve pledged to train 30 million individuals by 2030 in the skills that are needed for society today.

      It is probably the only thing to do. But the problem remains that even re-skilling want be sufficient if we will need less human labour.

    89. And so you see already people that are using GPT-4 to do their job much more efficiently by helping them with tasks. Now, GPT-4 will I think entirely automate away some jobs, and it will create new ones that we believe will be much better. This happens again, my understanding of the history of technology is one long technological revolution, not a bunch of different ones put together, but this has been continually happening. We, as our quality of life raises and as machines and tools that we create can help us live better lives the bar raises for what we do and, and our human ability and what we spend our time going after goes after more ambitious, more satisfying projects. So there will be an impact on jobs. We try to be very clear about that, and I think it will require partnership between the industry and government, but mostly action by government to figure out how we want to mitigate that. But I’m very optimistic about how great the jobs of the future will be.

      Fair statement. There is a bit naive view that we will increase happiness as we work less in some part of the work. But, so far, digital revolution has proven opposite. With Gig economy people work more and more. There is only sharp increase in inequality as capital becomes more relevant than labour.

    90. not a creature,

      God point on avoiding anthropomorphism.

    91. this technology is in its early stages

      As with Google and other tech companies, it is likely to remain in permanent 'beta version'.

    92. I think that’s a great idea.

      To be cynical - it is a 'great idea' because it won't work in practice, but there is pretention that we are doing something.

    93. The National Institutes of Standards and technology actually already has an AI accuracy test,

      It would be interesting to see how it works in practice. How can you judge accuracy if AI is about probability. It is not about certainty which is the first building block for accuracy.

    94. Some of us might characterize it more like a bomb in a China shop, not a bull.

      Q: Is AI bull or bomb in a China ship?

    95. Ultimately, we may need something like cern Global, international and neutral, but focused on AI safety rather than high energy physics.

      He probably thought of analogy with IPCC as supervisory space. But CERN could play role as place for research on AI and processing huge amount of data.

    96. But we also need independent scientists, not just so that we scientists can have a voice, but so that we can participate directly in addressing the problems in evaluating solutions.

      An important stakeholder.

    97. The sums of money at stake are mind boggling. Emissions drift, OpenAI’s original mission statement proclaimed our goal is to advance AI in the way that most is most likely to benefit humanity as a whole, unconstrained by a need to generate financial return. Seven years later, they’re largely beholden to Microsoft, embroiled in part an epic battle of search engines that routinely make things up.

      Why we should not trust AI companies?

    98. We want, for example, for our systems to be transparent, to protect our privacy, to be free of bias and above all else to be safe. But current systems are not in line with these values. Current systems are not transparent. They do not adequately protect our privacy, and they continue to perpetuate bias, and even their makers don’t entirely understand how they work. Most of all, we cannot remotely guarantee that they’re safe. And hope here is not enough. The big tech company’s preferred plan boils down to trust us. But why should we?

      What is the current situation in AI industry?

    99. We all more or less agrees on the values we would like for our AI systems to honor.

      Are we? Maybe in the USA, but not globally. Consult the work of Moral Machine which shows that different cultural contexts imply whom we would save in trolley experiment: young - elderly, man - women, rich - poor. See more: https://www.moralmachine.net/

    100. A law professor, for example, was accused by a chatbot of sexual harassment untrue. And it pointed to a Washington Post article that didn’t even exist. The more that that happens, the more that anybody can deny anything. As one prominent lawyer told me on Friday, defendants are starting to claim that plaintiffs are making up legitimate evidence. These sorts of allegations undermine the abilities of juries to decide what or who to believe and contribute to the undermining of democracy. Poor medical advice could have serious consequences to an open source large language model recently seems to have played a role in a person’s decision to take their own life. The large language model asked the human, if you wanted to die, why didn’t you do it earlier? And then followed up with, were you thinking of me? When you overdosed without ever referring the patient to the human health?

      Examples of risk narrative

    101. What criminals are gonna do here is to create counterfeit people.

      Risks narrative

    102. Choices about data sets that AI companies use will have enormous unseen influence. Those who choose the data will make the rules shaping society in subtle but powerful ways.

      What about each of us choosing datasets? AI has to be bottom-up.

    103. Fundamentally, these new systems are going to be destabilizing. They can and will create persuasive lies at a scale humanity has never seen before. Outsiders will use them to affect our elections, insiders to manipulate our markets and our political systems. Democracy itself is threatened. Chatbots will also clandestinely shape our opinions, potentially exceeding what social media can do.

      Risks narrative

    104. guardrails

      Guardrails are emerging lingo in AI governance.

    105. First, different rules for different risks. The strongest regulation should be applied to use cases with the greatest risks to people and society. Second, clearly defining risks. There must be clear guidance on AI uses or categories of AI supported activity that are inherently high risk. This common definition is key to enabling a clear understanding of what regulatory requirements will apply in different use cases and contexts. Third, be transparent. So AI shouldn’t be hidden. Consumers should know when they’re interacting with an AI system and that they have recourse to engage with a real person should they so desire. No person anywhere should be tricked into interacting with an AI system. And finally, showing the impact. For higher risk use cases, companies should be required to conduct impact assessments that show how their systems perform against tests for bias and other ways that they could potentially impact the public. And to attest that they’ve done so by following risk-based use case-specific approach.

      Q: What are 4 elements of precision regulation as proposed by IBM?

    106. a precision regulation

      Language

      Precision regulation is another concept to follow.

    107. a threshold of capabilities

      What is 'a threashold'. As always devil is in detail.

    108. We believe that the benefits of the tools we have deployed so far vastly outweigh the risks,

      Balancing narrative Opportunities 80 - Risks 20

    109. be My Eyes, used our new multimodal technology in GPT-4 to help visually impaired individuals navigate their environment.

      Optimistic narrative

    110. We think it can be a printing press moment.

      Paradigm shift narrative

    111. But the basic question we face is whether or not this issue of AI is a quantitative change in technology or a qualitative change.

      Critical question. It is quantiative shift which will evolve into qualitative one.

    112. We had four bills initially that were considered by this committee and what may be history in the making. We passed all four bills with unanimous roll calls, unanimous roll calls. I can’t remember another time when we’ve done that.

      Child safety online is one of the rare issues that unite all political forces worldwide. Will it be extended to AI?

    113. will we strike that balance between technological innovation and our ethical and moral responsibility to humanity, to liberty, to the freedom of this country?

      Balance narrative Choice narrative

    114. I was reminded of the psychologist and writer Carl Jung, who said at the beginning of the last century that our ability for technological innovation, our capacity for technological revolution, had far outpaced our ethical and moral ability to apply and harness the technology we developed.

      A good reminder of Jung's work. It is on the line of Frankenstein's warnings of Mary Shelly.

      ||Jovan||

    115. is it gonna be more like the atom bomb, huge technological breakthrough, but the consequences severe, terrible, continue to haunt us to this day

      Analogy - atomic bomb

    116. Is it gonna be like the printing press that diffused knowledge, power, and learning widely across the landscape that empowered, ordinary, everyday individuals that led to greater flourishing, that led above all two greater liberty?

      Analogy with Printing press

    117. We should not repeat our past mistakes, for example, Section 230

      Acknowledging msitake with Section 230.

    118. scorecards and nutrition labels

      Can this analogy work?

    119. known risks

      The real problem is in 'known'. We can deal with known knowns. (un)known unknowns are major problem.

    120. transparency

      AI principles

    121. Now we have the obligation to do it on AI before the threats and the risks become real. Sensible safeguards are not in opposition to innovation.
    122. we may need something like CERN, global, international, and neutral, but focused  on AI safety, rather than high-energy physics.  

      CERN for AI

    123. Chatbots can clandestinely  shape our opinions, in subtle yet potent ways, potentially exceeding what social media can do.  Choices about datasets may have enormous, unseen influence. 
    124. I call datocracy, the opposite of democracy:
    125. These  guardrails should be matched with meaningful steps by the business community  to do their part.
    126. an AI Ethics Board
    127. a lead AI ethics official
    128. on regulatory guardrails
    129. A risk based approach ensures that guardrails for AI apply to any application, even as  this new, potentially unforeseen developments in the technology occur, and that  those responsible for causing harm are held to account.
    130. a “precision regulation” approach to  artificial intelligence. This means establishing rules to govern the deployment of AI  in specific use-cases, not regulating the technology itself. 

      This is a new concept a 'precision regulation'

    131. international cooperation on AI safety, including examining potential intergovernmental oversight mechanisms and standard-setting. 

      Call for intergovernmental oversight

    132. safety standards, evaluation requirements, disclosure practices, and external validation mechanisms for AI systems subject to license or registration. 
    133. a combination of licensing or registration requirements
    134. adhere to an appropriate set of safety requirements,
    135. regulation of AI is essential,
    136. to minimize any harmful effects for workers and businesses.

      How?

    137. the Alignment Research Center (ARC)
    138. a Cybersecurity Working Group
    139. a Cybersecurity Grant Program
    140. novel security controls to help protect core model intellectual property.

      Security and intellectual property.

    141. We will continue to explore partnerships with industry and researchers, as well as with governments, that encompass the full disinformation lifecycle.