Addressing Bias and Discrimination - Achieving Diversity, Equity, Inclusion & Belonging (DEIB)

People miss a critical point: AI must learn from available data and in some types of AI from cases that are giving to the AI as "correct" cases. But no matter that, AI results have a chance (probability) of error. So, in my personal opinion, sometimes we are talking about AI like it would be a human being. Unfortunately it will jeopardize the use of AI which is a tool, no more than that.

Thank you, Sergio,
I appreciate your report. How do we encourage people to evaluate their data to ensure that it has the appropriate representation?

Jeffrey,
interesting question as it combines 2 topics: DEI and AI.
There is not much overlap. DEI is about humans and their values of respect and care. AI is about automating and learning from data.

If you had data about the DEI status of your organization, AI could produce some insights.

A very interesting case study about implementing DEI based on data can be found in HBR 09/2004 'Diversity as Strategy'.

Dear Jeffrey,

Ensuring Diversity, Equity, Inclusion, and Belonging (DEIB) within the AI space is a crucial concern, given the profound impact AI has on society. There are multiple stakeholders involved in this effort, ranging from tech companies and academic institutions to governmental bodies and non-profit organizations. Here’s an overview of the efforts and initiatives:

1. Tech Companies and Industry Leaders:
Inclusive AI Development: Many tech companies, including those at the forefront of AI like Google, Microsoft, IBM, and OpenAI, are focusing on inclusive AI development. This includes diversifying AI teams and integrating DEIB principles in AI design and deployment.
Bias Mitigation in AI Systems: Efforts are being made to identify and mitigate biases in AI algorithms, particularly those related to race, gender, and ethnicity.
2. Academic and Research Institutions:
Research on AI Ethics and Bias: Universities and research institutions are actively conducting studies on the ethical implications of AI and ways to address bias.
Educational Programs: These institutions are also focusing on creating more inclusive educational programs to nurture a diverse pool of AI talent.
3. Governmental and Regulatory Bodies:
Policy and Regulation: Governments are increasingly aware of the need for regulation in AI. Initiatives include developing policies that ensure AI technologies are used ethically and do not perpetuate discrimination.
Public Consultations and Advisory Committees: Many governments have set up advisory committees and public consultations to guide their AI strategies, with a focus on DEIB.
4. Non-Profit Organizations and Advocacy Groups:
Awareness and Advocacy: Organizations such as AI Now Institute, Data & Society, and others are actively working to raise awareness about the importance of DEIB in AI and advocate for responsible AI practices.
Research and Reports: These organizations often publish research and reports that highlight issues of bias and inequality in AI.
5. Standards and Frameworks:
Development of Ethical AI Frameworks: Groups like IEEE, the European Commission’s High-Level Expert Group on AI, and others are working on frameworks and guidelines to ensure ethical AI development.
6. Collaborations and Partnerships:
Cross-Sector Collaborations: There is a growing trend of collaborations between different sectors - tech companies, governments, academia, and civil society - to address DEIB in AI collectively.
7. Community and Grassroots Efforts:
Community-Led Initiatives: Grassroots organizations and community groups are playing a critical role in advocating for the inclusion of underrepresented voices in the AI conversation.
Challenges and Ongoing Efforts:
Continuous Monitoring and Improvement: Given the evolving nature of AI, continuous monitoring and improvement of DEIB practices are necessary.
Global Perspective: Efforts must also take into account the diversity of global perspectives, especially from regions and communities that have been traditionally underrepresented in tech.
Conclusion:
The effort to ensure DEIB in the AI space is multifaceted and ongoing, involving a broad range of stakeholders. While there is significant progress, the work is far from complete. Ensuring that AI is developed and deployed in an ethical, equitable, and inclusive manner remains a dynamic and critical challenge that requires sustained commitment from all sectors involved.

BR,

Markus

Jeffrey -

While there are no fool-proof safety nets for this, and it has been an issue even before the current generation of AI tools (see Cathy O'Neil's Weapons of Math Destruction for many older examples), there are a number of measures being designed or actively implemented including:

- Guidance from industry analysts and thinktanks
- Guardrails from regulatory bodies
- Internal organizational policies
- Active monitoring by either humans or different AI tools
- Training including coverage of what is and isn't appropriate usage

As far as representation goes, it is no different than any other change implemented through projects. If PMs and teams don't do a good job of identifying all the stakeholders who should be actively engaged (e.g. in providing fair, representative data for AI training purposes), then we will end up with gaps.

I particularly like this TED Talk as the speaker does provide a variety of options: https://www.ted.com/talks/max_tegmark_how_...i_under_control

Kiron

Thank you, Markus,
You provided a comprehensive response. What are your thoughts on how we can contribute to these efforts?