AI Ethics

Last updated — December 2023

J12 invests in companies that are taking advantage of the synergy between technological breakthroughs and market movements to create positive impacts on society and the environment, through either;:

Direct applications of AI systems, or
Indirectly enabling/enhancing the infrastructure, development, etc. of AI applications that produce positive impacts.

Being an actor within the space of Artificial Intelligence, given the technology’s substantial effects on the world and even more potential to be unlocked, we see the importance of meticulously defining what type of role we want to play, in what type of advancement of the overall field of stakeholders. Thus, by defining a set of AI principles, we are constructing an ethical perspective to be applied to AI systems, enabling a clearer identification of companies with inherently good use of technology that will win in the long term.

Our strategy is backing founders implementing relevant and moral AI ethics, as our bet is that they will build better companies, not only due to emerging legislation that will formalize such requirements, but also since ethical AI systems encompass better stakeholder alignment and lower risks.

AI Principles

Following emerging best practices, we have defined a set of principles to guide our operations and ensure a sustainable and socially beneficial course in navigating through the rapidly changing landscape of venture capital focused on AI and data startups.

We encourage our portfolio companies (and the stakeholders within J12’s own sphere of influence), as well as aim to inspire actors around us, to pursue ethical AI development and strive for the highest attainable sustainable and ethical nature of their AI systems they are directly or indirectly accountable for, in terms of:

Privacy
Respecting the Rights of Individuals

Clearly communicating to their users (and other relevant stakeholders) how their AI systems collect, store and use data.
Giving users control over how their data is collected, stored and used.
Communicating and giving control to users over their interactions with the AI systems.

Fairness
Mitigating Discrimination from AI Systems

Use unbiased data to create non-discriminatory AI systems.
Actively monitor and review data and models to avoid defects.
If bias is present, implement relevant de-biasing tools/methods .

Safety
Being Precautionary in Developing AI Systems

Conduct training of AI systems in constrained environments.
Set up safeguards during deployment.
Maintain explainable documentation of the AI systems.
Keep decision-making processes of AI systems human-centric.
Establish clear allocation of accountability with diverse groups of humans controlling the holistic goals and operations of AI development.

Regulatory Compliance

J12 is striving to be aligned with and contribute to advancing the forefront of optimally integrating AI into society and business.

We are closely following the development of the European Commission’s AI Act, striving towards not only being ready to adopt the new mandates for when the legislation is officially passed, but be among the leaders in applying and forming the best practices for AI development with regards to formalized regulation.

In waiting for the finalization of the relevant legislation, we aim at preparing our operations, and the structure and strategy of our portfolio companies, to facilitate a smooth adaptation to applicable laws and regulations. Additionally, we aim to emerge as an organization that is a source of useful insights into how contemporary implementation of AI regulation should be conducted in practice. This will primarily be done by J12 through one or several of the following methods:

Monitoring and documenting the risk levels of the AI systems developed by our portfolio companies (in relation to the European Commission’s system of risk classification).
Not investing in any AI systems that fall under the European Commission’s classification of unacceptable risk, encompassing AI systems with the potential to cause catastrophic impacts on society or the environment.
Ensuring that mandates of certification and registration of the associated portfolio company’s AI system(s) is correctly implemented.
Maintaining documentation of the nature and the effects of measures taken to establish/improve AI ethics (both independent innovation, and change induced by legal mandate).