How should a system of institutional control over the development of advanced artificial intelligence models and algorithms be built so that this development does not get out of hand and lead to negative consequences that are currently difficult to predict?
Should the development of artificial intelligence be subject to control? And if so, who should exercise this control? How should an institutional system for controlling the development of artificial intelligence applications be built?
Why are the creators of leading technology companies developing ICT, Internet technologies, Industry 4.0, including those developing artificial intelligence technologies, etc. now calling for the development of this technology to be periodically, deliberately slowed down, so that the development of artificial intelligence technology is fully under control and does not get out of hand?
To this, should the development of artificial intelligence be under control? - the answer is probably obvious, i.e. that it should. What remains debatable, however, is how the system of institutional control of the development of advanced artificial intelligence models and algorithms should be structured so that this development does not get out of control and lead to negative consequences that are currently difficult to foresee. Besides, if the question: should the development of artificial intelligence be controlled? - is answered in the affirmative, i.e. YES, then who should exercise this control? So, how should an institutional system of control over the development of advanced artificial intelligence models and algorithms and their applications be constructed, so that the potential and real future negative effects of dynamic and not fully controlled technological progress do not outweigh the positive ones. Well, at the end of March this year 2023, a number of new technology developers, artificial intelligence experts, besides businessmen, investors developing technology start-ups, including, among others, Apple co-founder Steve Wozniak and the founder or co-founder of such technology companies as PayPal, SpaceX, Tesla, Neuralink and the Boring Company, i.e. Elon Musk, Stability AI chief Emad Mostaque (maker of the Stable Diffusion image generator) and artificial intelligence researchers from Stanford University, Massachusetts Institute of Technology (MIT) and other AI universities and labs have called in a joint letter for at least a six-month pause in the development of artificial intelligence systems more capable than the GPT-4 published in March. The aforementioned letter acting as a kind of cautionary petition was published on the Future of Life Institute centre's website, advanced artificial intelligence could represent "a profound change in the history of life on Earth" and the development of this technology should be approached with caution. In this petition of sorts, there are warnings about the unpredictable consequences of the race to create ever more powerful models and complex algorithms that are key components of artificial intelligence technology. The aforementioned developers of leading technology companies suggest that the development of artificial intelligence should be slowed down temporarily, as the risk has now emerged that this development could slip out of human control. The aforementioned petition warns that an uncontrolled approach to AI development risks a deluge of disinformation, mass automation of work and even the replacement of humans by machines and a 'loss of control over civilisation'. In addition, the letter suggests that if the current rapid development of artificial intelligence algorithm systems gets out of hand, then the scale of disinformation on the Internet will increase significantly, the process of work automation already taking place will accelerate many times, which may lead to the loss of jobs for about 300 million people within the current decade and, as a consequence, may also lead to a kind of loss of human control over the development of civilisation. Developers of new technologies point out that advanced artificial intelligence algorithm systems should only be developed when the development of artificial intelligence is under full control, the effects of this development are positive and the potential risks are fully controllable. Developers of new technologies are calling for a temporary pause in the training of systems superior to OpenAI's recently released GPT-4 system, which, among other things, is capable of passing tests of various kinds at a level close to the best results passed by humans. The aforementioned letter also calls for the implementation of comprehensive government regulation and oversight of new models of advanced AI algorithms, so that the development of this technology does not overtake the creation of the necessary legal regulations.
In view of the above, I address the following question to the esteemed community of scientists and researchers:
Why do the creators of leading technology companies developing ICT, Internet technologies, Industry 4.0, including those developing artificial intelligence technologies, etc., now call for the development of this technology to be periodically, deliberately slowed down, so that the development of artificial intelligence technology takes place fully under control and does not get out of hand?
Should the development of artificial intelligence be controlled? And if so, who should exercise this control? How should an institutional control system for the development of artificial intelligence applications be built?
How should a system of institutional control of the development of advanced artificial intelligence models and algorithms be built, so that this development does not get out of control and lead to negative consequences that are currently difficult to foresee?
What do you think?
What is your opinion on the subject?
Please respond,
I invite you all to discuss,
Thank you very much,
Warm regards,
Dariusz Prokopowicz
The call for a pause in the development of advanced AI systems and the implementation of institutional control is due to the potential risks associated with uncontrolled AI development. These risks include the automation of jobs, the spread of disinformation, and the loss of control over civilization. The creators of leading technology companies recognize the need to approach AI development with caution and to ensure that the potential risks are fully controllable.
In terms of who should exercise control over AI development, this is a subject of ongoing debate. Some argue that government regulation is necessary to ensure that AI development is aligned with societal values and ethical considerations. Others argue that self-regulation within the tech industry is sufficient.
Building an institutional control system for AI development would require the collaboration of various stakeholders, including government regulators, industry leaders, and academic experts. Such a system would need to address issues such as ethical considerations, privacy concerns, and transparency in AI decision-making processes.
In conclusion, the development of AI should be subject to control to ensure that its potential risks are fully controllable. The exact nature of this control system, however, remains a subject of ongoing debate and will require collaboration between various stakeholders.
Dariusz Prokopowicz ,
The open letter[1] is wishful thinking at best. Let us sample some of the statements provided in the letter:
- "Powerful AI systems should be developed only once we are confident that their effects will be positive and their risks will be manageable."
The concept of manageable risk differs from individual to individual much less gather consensus on a population on what is a " manageable" risk
- "we call on all AI labs to immediately pause for at least 6 months the training of AI systems more powerful than GPT-4."
This is the most naive statement of the letter. What can be accomplished in 6 months? We have not even gathered consensus on the simple philosophical approach to ethics in the history of philosophy, much less on how to approach AI from an ethical standpoint. stopping AI research for 6 months will not have a significant impact (there are even journals dedicated to AI ethics in print tackling the issue for several years and there is pretense to solve this in 6 months!).
- "implement a set of shared safety protocols ...These protocols should ensure that systems adhering to them are safe beyond a reasonable doubt."
The concept of beyond reasonable doubt will eventually fall on what is reasonable practice in the field and this leads to weak safety protocols.
- "AI research and development should be refocused on making today's powerful, state-of-the-art systems more accurate, safe, interpretable, transparent, robust, aligned, trustworthy, and loyal."
This statement cannot be more ambiguous in term definition and the references of the letter do not help.
The concept of stopping AI research should not be confused with regulating the commercial use of the technology. The letter aims at the wrong issue. Research cannot be stopped for ML, since anyone with a reasonable amount of computing power (which you can buy anyway in amazon) can continue to work in the subject. But, regulating business is easier since the major players will have the greatest impact in slowing down the misuse of the technology. also by regulating business more targeted restrictions on its use can be applied rather than vague lists of research topics.
References
[1] https://futureoflife.org/open-letter/pause-giant-ai-experiments/
The question is very long and not stated succinctly. Same with the long winding explanation and background behind.
We are talking here, about "institutional control" and in general, this isn't good.
Institutional control sounds terrible, in fact, to the free mind and to researchers... Sounds like thing of the past....like non-democratic, dictatorship thing that takes you back to the dark eras when institutional control was EVERYTHING...
Why someone needs an institutional control on a thing that is practically controlled by people? Researchers create the AI, and they control it. There is no doubt about that.
The indirect question is, therefore, should be, do you need an Institutional Control over the researchers who are paid to create AI?
HOWEVER, MIND YOU, THAT ...
"Pull the plug off" and there will be no AI. :-)
This will send the "institutional control" to cycle the dynamos in order to create electricity so to control the creation of AI....
I think that these artificially imposed questions to impede innovation and create false threats within innovative researchers in the society.
Good luck with your questions, Dariusz Prokopowicz
Viara Van Raad ,
To address some of your points
a) Along the same lines, I could argue that the answer you have given is also not stated clearly.
b) While I do not agree with the open letter, I do disagree with the reckless use of ChatGPT by scientist who should know better. Your point on "Pull the plug off" is also wishful thinking at best and does not make a concise point. Since, anyone with a powerful enough laptop can implement LLAMA or other LLM. Transformers are also easy enough to configure and implement nowadays with the appropriate framework and some imagination.
c) It is not funny (as in the 'pull the plug' comment) how reckless release of LLMs and chatbots can lead to suicide[1] leaving two children without a father. Accountability of researchers and businesses is necessary. It is not a question of "impede innovation and create false threats" but of responsible use of a technology. If suicide is a false threats, then I am in the wrong planet.
d) It is also not funny how scientists cut and paste ChatGPTs answer without thinking how that answer was generated and if "real thought" went into it. Statistical patterns of LLMs are meant to be just that and are not equivalent to causation or logical reasoning. Scientist must learn to know better than this.
References
[1] https://www.euronews.com/next/2023/03/31/man-ends-his-life-after-an-ai-chatbot-encouraged-him-to-sacrifice-himself-to-stop-climate-
- Badges
- Science topic