Glen Robinson, the Chief Executive Officer at Platform One.
In the world of technological innovation, artificial intelligence (AI) emerges as a critical driver of growth, particularly within text analytics. This field—a specialized branch of natural language processing (NLP)—offers considerable opportunities for extracting insights from unstructured data. For business leaders in numerous industries, the allure of AI in text analytics is palpable, offering the potential to transform data analysis.
Based on my experience as a leader of a company offering an AI-based text analytics platform, this guide aims to provide new users and those considering investing in text analytic tools with a comprehensive overview of their AI governance considerations. Additionally, I offer ways to understand and overcome the related challenges in governing a dynamically evolving and socially challenging technology.
Start with an understanding of AI governance.
Governance in AI encompasses the policies, frameworks and ethical guidelines that ensure the responsible development and deployment of AI technologies. For text analytics, governance addresses critical aspects such as data privacy and security, ethical algorithm design and accountability. Those considering investing in such tools must prioritize governance as a central criterion in their decisions, recognizing its significance in mitigating risks and fostering sustainable growth.
The hallmark of a promising AI text analytics project is its commitment to robust governance practices. This includes adherence to ethical considerations, transparent algorithmic processes and compliance with privacy laws. But how do you know if your text analytics provider follows these practices? You can start by asking them the following questions.
Ethical Considerations And Bias Mitigation
• How do you ensure AI models are developed with ethical considerations in mind?
• What measures are taken to identify and mitigate biases in datasets and models?
Transparency And Explainability
• How do you enable transparency and explainability of your AI models?
• Are the decision-making processes of your AI models interpretable to lay users?
Data Privacy And Protection
• How do you enable compliance with global data protection regulations (e.g., GDPR and CCPA)?
• Can you explain your data anonymization and encryption techniques to a lay user?
Evaluate your team’s governance expertise and provide training.
Behind every successful AI project is a team that understands the importance of governance. When considering implementing an AI-based text analytics system, evaluate your team’s expertise in navigating the complex regulatory and ethical landscape of AI. A team well-versed in governance is better equipped to handle the challenges that arise during the development and deployment of text analytics solutions.
Following Platform One’s recent acquisition, a skilled team developing advanced AI faced challenges due to a lack of governance knowledge, leading to compliance issues, ethical concerns and security risks. This resulted in misaligned goals, wasted resources and stakeholder distrust. Recognizing the need for change, Platform One’s leadership initiated a comprehensive training program, incorporating workshops, expert seminars and mentorship to foster a culture of governance. Through practical exercises and case studies, the team gained a deeper understanding of their societal impact. Gradually, they became proficient in governance, enhancing the company’s resilience and trustworthiness. Their efforts led to compliant and ethically sound projects, establishing a new standard for AI development.
Keep up with the rapid changes.
The regulatory landscape and the foundational technology in text analytics are both subject to continual evolution. This dynamic environment poses a substantial risk of rendering existing products obsolete. Consequently, governance assumes a pivotal role in safeguarding the product’s relevance and viability. A robust governance framework must be adaptable and should be consistently applied throughout this process, serving as a guiding set of principles. However, the very nature of innovation is to push boundaries, and thus, it’s incumbent upon the governing body to anticipate and prepare for significant challenges and dilemmas that will inevitably test the resilience and flexibility of the system.
Both AI Now Institute and Partnership on AI (PAI) are coalitions committed to supporting the development and elevation of thought surrounding AI governance, so it’s a great idea to keep tabs on their websites. You could also subscribe to AI Now Institute’s free newsletter or consider becoming one of PAI’s partner organizations.
Guard against manipulated content and biases.
AI-generated deepfakes blur the line between reality and fabrication, posing challenges in distinguishing genuine from artificial content. Although beneficial for anonymizing video feedback to enhance client engagement, they risk being misused for misinformation, election manipulation or smearing individuals. It’s crucial for governing bodies to establish clear boundaries between freedom of expression, privacy and preventing harm from manipulated content.
AI systems, like those in hiring and lending, can reinforce societal biases from historical data, challenging fairness. Correcting these biases without reducing AI’s effectiveness or introducing new biases, as well as addressing inherent biases in development teams, is crucial. Governing bodies must regulate to ensure equity, acknowledging that legacy biases reflect certain worldviews within the developer and user communities.
Conclusion
AI intensifies ethical dilemmas, demanding societal discourse across regulatory, academic, industry and corporate spheres. Navigating these challenges requires vigilant governance, especially as AI in written communication presents unique ethical issues. Governing bodies must tackle these dilemmas, ready to question their own biases and reassess established norms.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
