Alliance experts participate in discussion on ethics in AI at OpenTalks.AI

February, 18th, experts from the AI Alliance spoke about the next steps in regulatory development and the most pressing ethical issues in the creation and application of artificial intelligence technologies at the Open Talks AI conference. The session also included representatives from companies that have joined the Code of Ethics.

Current ethical issues

“AI ethics is about ensuring that a consumer who uses a technology that is not always predictable is not left alone with possible negative consequences,” said Elena Suragina, one of the co-authors of the AI Code of Ethics and head of the regulatory support group for the MTS ecosystem, in her speech.

“I assume that over time there will be some kind of ethical standards that developers can rely on. They will secure them to some extent and inspire trust from the consumer,” she added.

Alexander Kraynov, director of artificial intelligence technology development at Yandex, illustrated the problem of ethics in the application of AI with a case study related to the ranking of search results.

“People have absolute confidence in search results: if a reputable search engine has ranked highly, it means it’s a good result. The challenge of responsible ranking emerges, but from an ethical perspective this is a huge challenge, because deciding what is good and what is bad is an incredibly dangerous way to go. This is one of the problems we have to solve. The general principle is clear: do not interfere anywhere except in matters that are matters of life and death, and in all cases where we do interfere, we should make this clear and explain the reasons,” Krainov said.

“I can guarantee that all newly introduced rules will be public and transparent, and we will explain why certain decisions are made, what targeting we give to the algorithm, what factors are used,” he stressed.

“One of the key tasks for us today is to create a list of the most frequently asked questions about AI ethics and suggest possible solutions,” commented Andrei Neznamov, managing director of Sberbank’s AI Regulatory Centre.

Regulation and self-regulation

The issues of state regulation were given a great deal of attention by the panelists. It was noted that the introduction of new regulations in this area is inevitable, it should be expected in the near future, and it is important to avoid a situation where this will hinder the development of technology.

“There are a certain number of activities in Russia and abroad aimed at regulation in the field of artificial intelligence. Some of them could become critical for the development of artificial intelligence, such as equating anonymised data with personal data. And this is really a big problem, because then, for the purposes of machine learning, impersonal datasets will need to be processed according to the appropriate rules,” said Elena Suraghina.
She noted that “now the community needs to draw up an action plan, to highlight the points which will become pain points in the next year.

Nikita Kardashin, head of Naumen’s integrated digitalisation practice, added: “We need to avoid a situation where restrictions will lead to a stagnation in technology development. There is a risk that over-regulation will lead to the fact that we at the state level will start to lag behind the world competitively, because our companies will suffer from this, while, let us say, Chinese companies will suffer from this to a lesser extent, and we will just get a big brake for ourselves.

“We understand that legislation will arise and regulate developers, so it makes sense to look closely at the Code of Ethics to avoid mistakes, to put certain restrictions on the use of data, on the use of methodologies so that our development is ethical and correct,” stressed Paul Drager, board member and vice president of software products and solutions at Sitronics Group.

According to him, “the Code can be a point of convergence between the interests of customers and implementers. That is a platform where people from business and IT-environment experts can speak the same language, operate with one conceptual apparatus and one set of constraints. This is an important point, as it should simplify life, both for vendors and customers, which is sometimes difficult to find a supplier with the desired level of reliability, especially when the budget is limited.

To join the Code, visit the AI Ethics Code section on AI Alliance Russia’s website.