The values behind Krino AI and ethical assessment
The Krino team is aware that the assessment list below will never be exhaustive. Ensuring Explainable and Responsible AI is not about ticking boxes, but about continuously identifying and implementing requirements, evaluating solutions, ensuring improved outcomes throughout the AI system’s lifecycle, and involving stakeholders in this.
The characteristics of Krino in relation to its design and building
The development, deploy and use of Krino adheres to the ethical principles of: respect for human autonomy, prevention of harm, fairness and explicability. The designers and builders acknowledge and address the potential tensions between these principles(HLG, Chapter I).
Krino designers and builders ensure that its development, deployment and use meet the seven key requirements for Trustworthy AI: (1) human agency and oversight, (2) technical robustness and safety, (3) privacy and data governance, (4) transparency, (5) diversity, non-discrimination and fairness, (6) environmental and societal well-being and (7) accountability (HLG, Chapter II).
Krino helps promote human autonomy as an aid to decision-making. It respects the goals of beneficial AI research, which should create not undirected intelligence, but intelligence beneficial to humans (AA and HLG, Chapter I). It will be as safe and secure as possible, and verifiably so where applicable and feasible (AA). If AI fails, it should be possible to ascertain why. This means that Krino does apply only algorithms that can be inspected by humans so to understand their steps (AA).
Krino should help preventing harm since it can be used for debunking fake news or finding critical errors in, e.g., safety systems. It is designed and operated so as to be compatible with ideals of human dignity, rights, freedoms, and cultural diversity (AA).
The application of Krino to personal data must not unreasonably curtail people’s real or perceived liberty (AA).
The power conferred to Krino by its designers and builders should respect and improve, rather than subvert, the social and civic processes on which the health of society depends (AA). The most significant risk is that Krino goes into the hands of malicious individuals and organizations, turning it into a machine that can create disinformation and frustrate its exposure as such. Krino designers and builders actively work to avoid this use.
Krino designers and builders acknowledge that, while bringing substantial benefits to individuals and society, AI systems also pose certain risks and may have a negative impact, including impacts which may be difficult to anticipate, identify or measure (e.g., on democracy, the rule of law and distributive justice, or on the human mind itself). They adopt adequate measures to mitigate these risks when appropriate, and proportionately to the magnitude of the risk (HLG, Chapter I).
Krino wil be developed and offered as open-source software, with optional closed-source “enterprise” components.
The characteristics of Krino in relation to its users and stakeholders
Krino should be beneficial and empower as many people as possible (AA). It promotes fairness and explicability in that any competent user can use it as a white-box AI.
The intention is that Krino will also be used by organizations whose aims and mission is to minimize inequality and discrimination, especially when it comes to the distribution of high-quality information.
Krino data owners should have the right to access, manage and control their data, given Krino’s power to analyze and utilize that data (AA).
Krino designers and builders should pay particular attention to situations involving more vulnerable groups such as children, persons with disabilities and others that have historically been disadvantaged or are at risk of exclusion, and to circumstances characterized by asymmetries of power or information, such as between employers and workers, or between businesses and consumers (HLG, Chapter I).
Krino addresses asymmetries of power or information in that it helps people to assess the quality and trustworthiness of information and to evaluate arguments on their merits.
A code of conduct for the Krino designers and builders
The Krino team applies a culture of cooperation, trust, and transparency among its members (AA).
The Krino team communicates, clearly and proactively, information to stakeholders about its capabilities and limitations, enabling realistic expectation setting, and about how the requirements are implemented. The Krino team will be transparent about the fact that stakeholders and users are dealing with an AI system.
Krino designers and builders are responsible for the moral implications of its use, misuse, and actions, with a responsibility and opportunity to shape those implications (AA). For this reason, they facilitate the traceability and auditability of Krino, and carefully select collaborations and use cases with particular attention in critical contexts or situations.
The Krino team fosters research and innovation to help assess AI systems and to further the achievement of the requirements. They will disseminate results in open access publication whenever possible for the community of AI scientists and practitioners as well as to the broader public.
The Krino team will consider technical and non-technical methods to ensure the implementation of the requirements, and the continuous assessment for a trustworthy AI (HLG, Chapter II and III), notably by:
- Making existing, potential, or new team members, as well as potential stakeholders and users, aware of the ethics guidelines provided in this document.
- Engaging in a continuous discussion with team members and (potential) stakeholders about fundamental tensions between different principles and requirements. This means continuously identifying, evaluating, documenting, and communicating these trade-offs and their solutions.
The Krino project involves a specific vision of entrepreneurial aspects.
We adopt the following values as the basis for professional collaborations within the Krino team and external partners:
We expect the same level of transparency among Krino customers, cooperation partners and external consultants as we hold our own organization, internal employees and collaborators to.
We adopt specific values in our collaborations, for instance:
- People are what is most valuable in a company.
- We strive for the best solution for our customers/Krino users, as that is best for our company
- We strive for technological innovation, but not at the cost of making at risk any team member.
- Human rights come first, immediately followed by customers’ interests. The Krino team and external collaborators will comply with the ethical code laid down in the first part of this document.
- Rules in the organization do exist, as minimal as possible, as many as necessary.
We implement the following entrepreneurial strategy:
- Any development and further tailoring of the Krino engine must be based on the core theoretical work of Gobbo, Wagemans, Russo.
- Open Source is required due to the necessity of
- publishing scientific research results
- proof of white-box approach: traceability, verifiability, inspectability
We adopt the following marketing strategy.
We aim to use what is widely known as the Red Hat business model. This business model is constituted by the following approach:
- A professional open source publishing policy, touching all our scientific results and the core of our software, including such documents as benchmarks and documentation
- Professional support, maintenance and customisation on a paid basis; support could e.g. be provided subject to a subscription or renewable license.
- Possible “enterprise” versions of our software under a paid license.
- Workshops, courses and consultancy around the software on a paid basis.
- Coupling of Krino to the Sparx Systems Enterprise Architect software tool, via one or more of that software’s APIs, thus creating considerable added value for existing Sparx Systems customers.
We adopt the following provisions for risks mitigation
- Avoid nontransparent partners
- In the first instance, focus on engineering use cases – the domain where the engineering team of Krino has the best expertise and certainty; in the second instance, enlarge scope and application of Krino to other use cases.
- In any potential cooperation with the military, secret services, or political parties, where sensitive use cases may be at stake, make sure that all of our ethical codes detailed in section 1 are met.