It’s been years since Artificial Intelligence became part of our everyday language. Yet, even today, we refer to this technology as humanizing it. The question “is Artificial Intelligence ethical?” is a poorly posed question because it assumes that the technology has consciousness. Instead, it is a tool whose effects depend solely on the programming done by humans. However, understanding what design boundaries should not be crossed helps limit its risks. Above all, when using personal or sensitive data circulating on the net.
Algorithms and privacy: what the limits should be
The concept of privacy has never been as debated and relevant as it is now. The reason is simple: a huge amount of data is produced in the digital world. Not all of it is functional to the activities we want to digitize; a lot of it is related, for example, to personalizing the customer journey to obtain a good customer experience. Let me explain.
It is important to know our target audience to sell a product or service. Initially, it was enough for us to study consumption behavior at the point of sale, but not anymore. With Artificial Intelligence embedded in technologies and digital media, it is possible to collect data, for example, on the entire purchase path of a potential consumer. Knowing what they do when they connect to our e-commerce, which products they prefer, and which Apps they use for payments.
How do companies collect this data? Through the consent expressed when we accept the banners present on the sites we visit. Between the lines, it says that our data will be tracked. In our haste to access site content, we often allow our data to be processed without reading the specifics. Every mouse click on the site we are using will be detected and recorded from that moment.
Whether this behavior is ethical or not certainly does not depend on the technology or software used. Moreover, the company’s responsibility is only to ensure maximum transparency in data collection and management. It is we, as users, who must pay attention.
The challenge for institutions like the European Commission and the Federal Trade Commission is precisely to protect people from the negative effects of using their personal information. The attack, which takes place through stringent laws on the use of privacy, is aimed at companies and programmers and not at Artificial Intelligence.
What might be some unethical uses of AI?
When we have a powerful tool like Artificial Intelligence in our hands, the risk of using it for unethical purposes that harm other people is very strong. In fact, not always business managers evaluate the possible ethical implications before programming an AI algorithm.
This leads to unethical uses that undermine corporate credibility. Such misuses can occur in various areas:
- AI learning can be used by cybercriminals to recreate automated email templates similar to the real thing. For example, in the body of the email, the text picks up phrases used by the CEO in the same style so as to make recognition impossible. In these cases, it can save us from knowing how to protect our digital identity.
- Another unethical use of AI is to spread propaganda and misinformation. There are online movements that, especially through social media, hook followers through fake news and misleading and inaccurate images. Or real videos with faces modified by AI, as in the case of Deepfake.
- Employing AI algorithms for recruitment is not entirely unethical but can become so because of the input data. To train the algorithms, previous data is used, which often reflects a very different job landscape than we have today. In the digital age, the professions and skills required are constantly evolving. Therefore, it happens that the algorithm becomes discriminatory and unethical due to a lack of up-to-date data that can include all the variables needed to ensure fair treatment.
Ethically designing an algorithm with Artificial Intelligence means promoting improved and equitable goals for people and society Click To Tweet
It is therefore important to reflect on how decisions made in AI programming can have a broad ethical and social impact. The book ‘Toward a Post-Digital Society: Where Digital Evolution Meets People’s Revolution’ offers an in-depth analysis of these topics, focusing on the role that emerging technologies will play in our future.
What might be some ethical uses of Artificial Intelligence?
Programming software that positively impacts society means pursuing goals to improve people’s private and professional lives.
Here are some uses that ethically impact society:
- AI used in medical settings to save lives. In this case, learning, for example, allows to speed up the time of diagnosis of a disease. Only such powerful software, in fact, can quickly scan and combine a huge amount of data from all over the world.
- Through custom-designed programs, algorithms are used to improve children’s learning with cognitive-behavioral difficulties. In this case, the algorithm adapts to the children’s abilities and improvements.
- Intelligent crop monitoring is another ethical use of AI that allows large areas of crops to be monitored using computer vision. This use allows, for example, monitoring to prevent damage or fires that could compromise yields.
How to design ethical AI software?
When you decide to implement software equipped with Artificial Intelligence in your company, the most delicate phase is the design phase. Both because it impacts the results we want to obtain; and because it is in this step, we risk colliding with the ethical values of those involved.
Suppose the organization wants to create a machine learning algorithm that is economically viable but at the same time preserves the autonomy of people. In that case, it can design based on ethical principles.
What you need to consider when designing software with Artificial Intelligence:
- First, pay close attention to the input data on which the training will be based. It is necessary to evaluate the origin of the data, which must be reliable, and the interferences suffered during the processing of the data;
- Use a sufficiently large volume of data that considers the variables and all possible scenarios of the contexts in which they will be implemented.
- Aim for goals that promote the well-being of people and the planet. In general, the purpose of AI should be to provide resource support, reduce inequalities, and ensure fair treatment.
- Train resources to have a thorough understanding of the technology and understand the ethical implications that its application might generate.
- Evaluate the impact of the algorithm. The control system and the definition of responsibilities are fundamental steps to limit the risks generated unknowingly.
The real problem is that often, those who design algorithms fail to be objective. There is always a veil of prejudices rooted in our society in their choices. The important thing is to pay attention and, if necessary, modify the models at the first hint of ethical or moral harm. AI designers should ensure that the use of such technology pursues values such as respect, fairness, and transparency.