Photo credit: tamasmatusik on Visualhunt / CC BY-NC-ND
* Andriei Gutierrez
The world is questioning how the authorities should treat Artificial Intelligence (AI). Should AI be guided by some minimum parameters? Should regulation come a posteriori, if necessary? Or should it already be regulated? Many countries, multilateral organizations, the specialized press, academia, business and civil society organizations are concerned about these issues.
In this broad debate, I highlight the recent EU Ethical Guidelines for reliable and secure artificial intelligence and the recommendations of the OECD Council on AI, to which Brazil was a signatory. These are pioneering and inspiring initiatives that must be known and studied carefully, as they bring ethical parameters and recommendations of relevant public policies for the advancement of AI in Brazil.
On the other hand, these discussions have not yet received due attention here in Brazil. With rare exceptions, we still have few debates, events, articles or proposals for public policies on the subject. In this context, I decided to share 6 premises that I consider essential to advance the debate on AI in the country.
1. AI innovation needs to be supported by a risk-based approach combined with the establishment of ethical principles and values. Current rapid technological advances have taught us that nations that support innovative ideas in a risk-based approach are more likely to succeed. In this sense, it is understood that the potential benefits outweigh the risks and it allows innovations to flourish combined with mechanisms to mitigate them.
On the other hand, the necessary incentive to AI-based innovation does not mean the total and absolute absence of minimum principles and conditions for the development and application of AI-based solutions. The international debate has moved towards the realization of the need for such minimal goals, so that AI-based systems are created and executed in a way that respects human life and excels for their well-being, diversity and inclusion, and democratic principles. and privacy, in addition to paying special attention to security, explainability, transparency, control and centrality in human beings.
In this field, it would be very pertinent to make serious progress towards the recommendations made by the OECD previously mentioned.
2. We don't fall for simplistic temptations, like a general regulation for AI. At IBM, we have had the opportunity to help organizations from more than 20 sectors in their digital transformation, of which the adoption of AI is an integral part. From this experience, I am increasingly convinced that any limits and parameters for the innovations and uses of AI systems must be very well discussed and delimited in their different sectors, in a timely and very precise manner. Otherwise, we will create a restraint on innovation and the adoption of AI in Brazil with harmful consequences for the present economic and social development and for future generations.
3. AI investments need progressive legal certainty. On the one hand, it would be pertinent to legal certainty that we quickly define situations in which the use of autonomous devices or systems will not be allowed and when it will need the supervision or tutelage of a human being. As we do not yet have a crystal ball capable of predicting all the possible implications of AI, a global set of principles would have the merit of bringing this important guideline and guaranteeing the necessary legal security for companies, governments and research institutions. This security is essential to encourage large-scale AI adoption and advancement.
If edges to be trimmed and paths to be indicated still persist, a timely and precise debate on possible regulations, whenever possible, at the infra-legal level is pertinent.
4. The decision on the limits for the use of AI should be based on technical arguments, but ultimately it will be a political decision that will require a wide debate. Which criteria should prevail when assessing and deciding whether a robot can perform surgery without human supervision, whether we will accept the development of civilian or military autonomous weapons, or whether it will be possible to transport passengers and cargo by autonomous vehicles and aircraft? What will be the legal and blameworthy implications in the event of intent?
I am increasingly aware that, although this decision can and should be supported by technical criteria, be they social, economic, environmental and ethical, everything indicates that the decisive instance will be politics. And there is still a lot of water to run in that river; much debate to be done. It is necessary that the debate on AI and public policies for AI leave the technical spheres, today concentrated in organizations in the technology sector, a few specialized government agencies, academics and specialists in the subject. It is necessary for Brazilian society to take ownership of the theme and make a qualified debate about the importance of AI for the present and the future, its impacts and the public policies necessary for its sustainable advancement.
5. The risk-based approach must be accompanied by mitigation and accountability (digital) mechanisms. The speed and scale of AI-based solutions and services will require us to rely increasingly on digital tools - and based on AI itself - to monitor, audit and ensure their accountability. In other words, the digital society will increasingly demand that we use digital tools to mitigate risks of a digital nature. Likewise, we need to move towards a regulatory model that provides, whenever possible and pertinent, for the establishment of procedures and the use of such types of tools.
6. AI will bring about such profound social transformations that the country needs a national strategy to prepare. Important issues demand debates and (above all) actions coordinated by the country. It is important that Brazilian society moves at the same pace and direction with public policies for the different relevant domains of AI for the country, in applied research, in stimulating the adoption of AI by the public and private sector, in the development of digital skills of the hand of work, in the progress towards the adoption of global and interoperable standards, in addition to the necessary security, trust and the establishment of ethical principles.
We will work for that.
* Andriei Gutierrez is the Coordinator of the Regulatory Committee of ABES and of the Movimento Brasil, País Digital - https://brasilpaisdigital.com.br
EN 
PT 
