Experts state that artificial intelligence is a very significant power and its sphere of influence is rapidly expanding, arguing that the production and use of such a power, if concentrated in a single hand, could alter global balances. Highlighting that the era requiring the control of artificial intelligence is approaching, Dr. Öğr. Üyesi Nuri Bingöl emphasizes that this technology could carry risks equivalent to nuclear energy if adequate precautions are not taken. Bingöl, noting that there is no answer yet to the question of how artificial intelligence supervision will be implemented, states that rather than a global authority, each country should have its own legislation and regulations.
Üsküdar Üniversitesi Faculty of Health Sciences Vice Dean, Head of the Artificial Intelligence Engineering Master's Program Dr. Öğr. Üyesi Nuri Bingöl evaluated the potential risks that artificial intelligence models could bring and how these risks can be controlled.
Artificial intelligence could take away many people's jobs across a large domain
Stating that omnipotent artificial intelligence models also bring potential risks, Dr. Öğr. Üyesi Nuri Bingöl said, “Due to the great advantages they provide to their users, it is currently unnecessary to consider sensational risks such as these individuals taking humanity into their grasp. However, many risks and dangers whose effects we need to confront much sooner await us.”
Reminding that with the industrial revolution, steam engines took over jobs previously done by people but also created many more job opportunities, Bingöl stated, “Although it is said that artificial intelligence will ‘create new job opportunities,’ the general models that have recently attracted the attention of people from all walks of life were not designed to perform human tasks more efficiently, like steam engines. They were designed to perform all kinds of cognitive activities that humans can do. This means that everything we see as a new job opportunity could be learned again by general models after a while. Businesses could emerge that operate without any human involvement in the process. If not managed correctly, the impacts of such a change could be very devastating globally. General-purpose artificial intelligence could take away the jobs of many people across a large domain, from medicine to law, from software engineering to management. Furthermore, they could work 24/7 and create a much cheaper cost than human colleagues.”
The concentration of such a significant power in a single hand could alter global balances
Drawing attention to the need to consider the risks brought by centralized supervision for ‘model creation and development,’ another stage of artificial intelligence, Bingöl said that it being done by a single authority poses a significant risk.
Stating that artificial intelligence is a very significant power and its sphere of influence is rapidly increasing, Bingöl said, “The production and use of such a significant power, whether global or not, if concentrated in a single hand, could alter global balances. Today, the problems created by nuclear power and its role in the global balance of power are evident.”
If adequate precautions are not taken, artificial intelligence technology could carry risks equivalent to nuclear energy
Referring to an article published by company founders and executives who have made significant advancements in artificial intelligence technologies, Dr. Öğr. Üyesi Nuri Bingöl said, “The article states that artificial intelligence technologies must be used responsibly and supervised correctly. It says that if this technology is not supervised, it could carry risks that might create negative impacts on human life and society. According to the article, the era requiring artificial intelligence to be controlled by international institutions is approaching, and if adequate precautions are not taken, this technology could carry risks equivalent to nuclear energy.”
Transforming global power into a community of powers raises other questions
“If we proceed from this analogy of nuclear power and risk, it is clear what consequences the concentration of artificial intelligence technology, globally or not, in a single hand will bring and how it will affect the global balance of power,” said Bingöl, continuing his words:
“To reduce this risk, transforming global power into a community of powers with multiple actors raises other questions. How will the elements representing global power be determined? To what processes and authorities will the use of this power be tied? How will disputes be resolved, and how will approval and veto rights be managed fairly?”
If not managed well, artificial intelligence could lead to more problems and power imbalances
Giving an example from the work of the World Health Organization during the pandemic, Bingöl said, “The World Health Organization is a highly respected and accepted organization affiliated with the United Nations, conducting international work on public health. Despite this, to what extent was it able to contribute to all communities worldwide benefiting from emergency health services during the COVID-19 pandemic? If states holding vaccine development power during the pandemic influenced the global balance, which power used this and how? As seen, the global authority model we expect to be a solution to artificial intelligence supervision and limitations could not remedy all the problems of the models we have experienced in other areas today. It is evident that a technology as powerful as artificial intelligence, if not managed well, could lead to more problems and power imbalances.”
Every stage of production and use can be spread across many different and independent authorities
Stating that artificial intelligence supervision and limitations do not necessarily have to be ‘circuit breaker’ like structures tied to a single center held by a global power, Bingöl said, “They can be thought of as open and transparent regulable regulations, laws, and legislations, including various standards, frameworks for operation and audit trails, and authority supervision infrastructures. In this regard, the roles and responsibilities of various actors can be regulated by laws and legislations. By distributing these powers, multilateral independent controls can be ensured. Intervention policies to be followed in situations of operation that get out of control, creating threats and dangers, can be determined. These policies can be pre-rehearsed and audited, similar to cybersecurity procedures. Every stage of production and use can be spread across many different and independent authorities. Instead of supervising processes like the creation and training of models with an international authority, after a model is created, the question of ‘how does it affect people’s lives and does it violate human rights?’ should be supervised.”
No obstacle should be created for institutions needing AI capacity that does not pose a national threat
Stating that like every technological development, artificial intelligence technology also needs resources, commercial models, and interest for its development, Dr. Öğr. Üyesi Nuri Bingöl said, “The supervision and limitations to be created regarding artificial intelligence will bring additional significant burdens. New compliance processes, obligations, process financing, and resource usage limitations… However, no obstacle should be created for institutions that need artificial intelligence capacity to a degree that does not pose an existential or national threat to humanity.” He also drew attention to the question of who will determine the boundaries that will create this expectation and by what criteria.
There is no answer yet to the question of ‘how’ supervision will be implemented
Stating that there is still a long way to go regarding how supervision and limitations will be implemented, Bingöl said, “For this, it is inevitable for government institutions, technology producers, and thought leaders on this issue to work together and produce solutions. Especially, ‘how’ these solutions will be implemented without hindering development is a subject of study and there is no answer yet. The idea of forming an international authority will only make sense when this answer is found. The answers to questions such as how the powers of this international authority will align with individual, national, and global power-holding companies, and what supervision mechanisms and powers it will use, will also become clear.”
Every country should have its own legislation and regulations
Dr. Öğr. Üyesi Nuri Bingöl stated that rather than a global authority, each country or union of countries should have its own adaptable legislation and regulations, while adhering to fundamental principles and compliance, saying, “Building a balance based on mutual obligations eliminates significant risks for our country and brings advantages in return.”
Reminding that Turkey is subject to many international organizations and regulations, and that it has complied with the obligations they bring without compromising its independence and security, and has been able to fulfill its commitments, Bingöl concluded his words as follows:
“Our country has the necessary official, public, and private institutions, universities, R&D centers, and civil society organizations to ensure this. It has the experience and capacity to create legislation that will ensure coordination among them. In the global arena, Turkey must also be at the center of such organizations. It deserves this situation by producing, using, and developing highly advanced artificial intelligence technologies in recent years to maintain the balance of power and have a say. Furthermore, it should take on a pioneering role through artificial intelligence in matters of justice, rights, and law.”