A report A recent report about the use of remote workers in India by Amazon to ensure the operation of the “Just Walk Out” system (or “just go out”, in free translation) brought up crucial questions about Ethics in Artificial Intelligence.
In theory, computer vision technology promised, according to the company, to automate the entire purchasing process, without any human interaction. All you had to do was enter the store, add the products to the cart and leave. The charge would be made automatically. However, after the controversy, the company said there was human oversight “for validation only”. But it's a fact that we'll never know for sure.
This is just one case of many that we may have questions regarding the use of AI. For that reason alone, many governments are working to regulate the use of this technology and define concepts such as explicability, ethics, and bias, with the objective of promoting fairer AI and - why not? - “free from controversy”.
The path to more ethical and transparent AI
This story about Amazon ended up bothering me profoundly. The thing is that, like other technologies that we developed here at Venturus, there have already been questions from customers about the possibility of a similar solution, which would automate the entire business process. But how do you know how accurate or effective Amazon's solution was?
As a technology professional, I have participated in several innovative advances over the years and witnessed the rise of AI - something that is there for all to see now. This case raises concerns about transparency and accountability in the development and use of this technology.
That's where the idea of Explanability as an imminent legal need, requiring transparency and accountability in AI algorithms. But what does that mean?
According to professors Cristiano Colombo and Wilson Engelmann, explicability can be understood as guarantee that mechanisms will be developed to ensure the transparency of the algorithms that make up the AI systems. For them, this requirement will be as much a trend as LLM, Generative AI, and other terms that we're getting used to hearing.
When ethics enter technology
In a broader concept, ethics in artificial intelligence refers to the development and application of AI systems in a way that is Morally correct And that respect fundamental ethical values.
This includes ensuring that AI does not perpetuate prejudice or discrimination (so-called “bias”), is transparent in its operations and decisions (explicability), and that its implications and uses are considered in terms of social and moral impact.
In addition, Ethical AI also involves questions about data privacy, consent to the use of personal information and responsibility for actions and decisions taken by AI systems.
The bias in artificial intelligence
Bias was one of the first controversial points raised by AI algorithms. This topic began to gain attention in the last decade, especially when AI became more integrated into everyday life and important decision-making processes.
The recognition that algorithms can perpetuate or even amplify prejudices existing social and racial issues emerged with notable cases and academic studies, leading to a growing awareness of the importance of ethics and justice in AI.
Concerns focus on how the data used to train these algorithms may contain historical or cultural biases, thus influencing AI behavior and decisions.
Researcher Joy Buolamwini was one of the pioneers in this field, and in and of itself, it is a subject that includes several articles. But there is already an excellent documentary called”Coded Bias” who tells that story very well!
The Brazilian Artificial Intelligence Strategy (EBIA)
A Brazilian Artificial Intelligence Strategy, of which Venturus is a part, brought much of this debate, according to the Final Report of Axis 6, on Research, Development and Innovation.
In the document, the EBIA proposes about 11 actions to define ethical, safe, and regulated AI. Among them, the use of Living Labs Regulatory, in which you can test, in the reality of organizations, the models of regulated self-regulation structured on the basis of principles.
Another important point is to have more technological exchanges so that we can have in Brazil the adoption of principles that are now being established in Europe, China and the USA, for example. Most of all, there is a reflection raised in the working groups regarding the Regulation of AI: O The use of AI has to be ethical!