Ethical Artificial Intelligence is at the heart of Omina Technologies, an AI and machine learning company. In June 2019, Rachel Alexander, the founder, and CEO won the Belgian AI Personality of the Year Award. “Getting the award, for me, is a real validation of ethical AI. It is crucial to have people in your organization who believe in ethical AI because it is a mindset that you need to develop. When I started our business in 2016, I had to explain the dangers of AI. Now, I focus on how we can take concrete steps to solve it.”
AI in Europe
On April 8, 2019, the independent High-Level Expert Group on AI set up by the European Commission presented the Ethics Guidelines for Trustworthy Artificial Intelligence.
The guidelines put forward a set of seven critical requirements for trustworthy AI systems. The group also prepared a document which elaborates on a Definition of Artificial Intelligence used for the guidelines:
Artificial intelligence refers to systems that display intelligent behavior by analyzing their environment and taking actions – with some degree of autonomy – to achieve specific goals. AI-based systems can be purely software-based, acting in the virtual world (e.g., voice assistants, image analysis software, search engines, speech and face recognition systems) or AI can be embedded in hardware devices (e.g., advanced robots, autonomous cars, drones or Internet of Things applications).
“I find it encouraging that the European Commission came out with a definition of AI and seven key requirements that ethical AI systems have to meet, says Rachel.”
Seven requirements for trustworthy AI
Requirement 1: Human agency and oversight
AI systems should empower human beings, allowing them to make informed decisions and fostering their fundamental rights. At the same time, proper oversight mechanisms need to be ensured, which can be achieved through human-in-the-loop, human-on-the-loop, and human-in-command approaches.
“This requirement points out that it is vital to combine AI and human intelligence. In other words, AI is augmenting human intelligence. There are studies about the diagnosis’s accuracy done by a doctor, done by AI, and done by both working together. The outcome shows that AI plus human intelligence is more accurate compared to only AI or human intelligence,” explains Rachel.
Requirement 2: Technical robustness and safety
AI systems need to be resilient and secure. They need to be safe, ensuring a fall back plan in case something goes wrong, as well as being accurate, reliable, and reproducible. That is the only way to ensure that also unintentional harm can be minimized and prevented.
“It is necessary that you have certain governance around critical AI systems to make sure that it is not possible to reverse engineer it. Additionally, how do you know that your AI system is accurate, reliable, and reproducible? To answer this question, you have to understand your algorithm thoroughly.”
Requirement 3: Privacy and data governance
Besides ensuring full respect for privacy and data protection, adequate data governance mechanisms must also be ensured, taking into account the quality and integrity of the data, and ensuring legitimized access to data.
“Since the General Data Protection Regulation (GDPR) came into effect on May 25, 2018, organizations focus more on privacy and data governance. However, I expect this to change in the future. Why? Because many youngsters don’t have a strong belief in data privacy due to, e.g., Facebook and voice assistants like Alexa.”
Requirement 4: Transparency
The data, system, and AI business models should be transparent. Traceability mechanisms can help achieving this. Moreover, AI systems and their decisions should be explained in a manner adapted to the stakeholder concerned. Humans need to be aware that they are interacting with an AI system, and must be informed of the system’s capabilities and limitations.
“Monitoring the evolution of your algorithms, and understanding what those are doing is an essential level of transparency. At our company, for example, we make it possible for customers to generate an output report by pressing a simple button. It shows you immediately if the AI system meets its service level agreements, what the risks are in the decision making of the algorithm, and how this has evolved.”
“It is crucial to have people in your organization who believe in ethical AI because it is a mindset that you need to develop.”
Requirement 5: Diversity, non-discrimination, and fairness
Unfair bias must be avoided, as it could have multiple negative implications, from the marginalization of vulnerable groups to the exacerbation of prejudice and discrimination. Fostering diversity, AI systems should be accessible to all, regardless of any disability, and involve relevant stakeholders throughout their entire life circle.
“I hear many people talking about getting the bias out of AI. That is impossible because you can’t get the bias fully out of humans and data. Nevertheless, what you can do is to minimize it by making it visible through the transparency of your AI system.”
Requirement 6: Societal and environmental well-being
AI systems should benefit all human beings, including future generations. It must hence, be ensured that they are sustainable and environmentally friendly. Moreover, they should take into account the environment, including other living beings, and their social and societal impact, should be carefully considered.
“We make our ethical AI and machine learning platform available for enterprises with lots of data and small and medium businesses with fewer data, tells Rachel.”
Requirement 7: Accountability
Mechanisms should be put in place to ensure responsibility and accountability for AI systems and their outcomes. Auditability, which enables the assessment of algorithms, data, and design processes, plays a key role therein, especially in critical applications. Moreover, adequate an accessible redress should be ensured.
“To assess if an AI system is ethical, you have to audit your AI systems and be transparent about the results and improvement plans. Overall, it is impossible to solve all ethical AI problems outlined in the seven requirements at once. Currently, our business focuses on privacy and data governance, transparency, and societal and environmental well-being, says Rachel.”
Three AI and machine learning offerings
Omina Technologies offers three AI services: Omina Academy, Omina Consultancy, and Omina Core.
The academy focuses on bringing AI knowledge to the market so that you can explore what it can do for your company.
Rachel: “Besides the academy, we also provide consultancy. Organizations often contact us with a request to help them understand how they can strategically implement AI. So, we come in and start with a readiness assessment. For example, we perform a benchmarking of the customer’s sector, assess the organization’s strengths and weaknesses to implement AI, and look at their data. That results in a strategic AI roadmap.”
“Last but not least, Omina’s Core is an ethical AI and machine learning platform for enterprises and small and medium businesses. It analyzes massive amounts of data, structured and unstructured, and makes complex decisions and predictions faster and more reliable.”
Do you want to enhance your AI knowledge and apply ethical AI and machine learning in your company? Contact: Rachel.Alexander@ominatechnologies.com