Trust [article]

Britain dislodged it conservatives. After fourteen years, the Labour Party stormed back to power. For our ruling party, 14 years will be one before the next general election and there could be a dislodge here too.

In his first speech as Prime Minister on 5-Jul-2024, Keir Starmer directly acknowledged the lack of trust between the public and politicians, stating that "this wound... this lack of trust... can only be healed by actions not words."

Starmer recognized that the "gap between the sacrifices made by people and the service they receive from politicians" had led to a "weariness in the heart of a nation" and a "draining away of the hope, the spirit, the belief in a better future." He pledged that his government would work to restore this trust, saying "we can make a start today with the simple acknowledgment that public service is a privilege and that your government should treat every single person in this country with respect."

The Prime Minister emphasized that rebuilding trust would be a key priority for his government, stating that "politics can be a force for good - we will show that. And that is how we will govern. Country first, party second." He promised to focus on delivering change and tangible improvements for the public, rather than empty rhetoric, in order to regain their trust.

By emphasizing trust, Starmer acknowledged its central role in effective governance and public confidence.

Trust forms the bedrock of political legitimacy and is essential for implementing policies and fostering social cohesion. Trust is the glue that holds relationships together. When we trust someone, we are willing to be vulnerable, share our thoughts and feelings, and rely on them to act in our best interests. If leaders have to build trust amongst people in the public sphere and people still have trust issues amongst themselves, how will they trust AI?

The challenges of establishing trust in AI systems were explored in a Harvard Business Review article from May 2024, "AI's Trust Problem." The article introduced the concept of the "AI trust gap," which is closed when individuals are willing to entrust machines with tasks typically performed by qualified humans. This gap represents the hurdle that AI must overcome to be fully integrated into various aspects of our lives.

The article outlines several prominent concerns that fuel skepticism and make it difficult for AI to earn trust, including:
- Misinformation
- Safety and security issues
- The opaque nature of AI systems
- Ethical dilemmas
- Biases
- Unpredictability
- Hallucinations in large language models
- Unforeseen risks
- Potential job displacements and societal disparities
- Environmental impacts
- Market dominance
- Governmental intrusions

The article suggests that a consistent strategy to address the "AI trust gap" involves:
1. Educating the public about AI
2. Empowering humans to be involved in the management of AI technologies
3. Ensuring human oversight and control over critical AI applications

By taking these steps, the article argues, the trust gap can be closed when people are willing to entrust machines to do jobs that would otherwise be done by qualified humans.

The rapid advancement of technology, particularly artificial intelligence (AI), has introduced new dimensions to the concept of trust. A Forbes article from July 3, 2024, titled "We'll Never Fully Trust Artificial Intelligence With Our Businesses — And That's Okay," also touches on this issue, arguing that while AI can be a valuable tool, it is unlikely that humans will ever fully trust AI with their businesses or personal decisions. This is because trust is inherently a human-to-human interaction, and AI lacks the emotional and social intelligence that humans possess.

Joe McKendrick's key points are:
Human Oversight is Essential: AI is unlikely to operate autonomously for complex tasks, and human intervention is necessary to scrutinize and potentially correct AI-generated outcomes. Symbiotic Relationship: AI should complement human expertise rather than replace it entirely. This collaboration is crucial for achieving optimal results.
Gradual Autonomy: As AI is integrated into various applications, there is a gradual shift towards granting AI more autonomy based on the complexity of operations.
Risk Management: In scenarios where AI significantly impacts human lives and rights, humans must assess risks and determine the appropriate level of oversight before deploying AI-driven services or products.
Monitoring Mechanisms: AI can assist in facilitating human oversight by integrating monitoring mechanisms to track AI actions and behaviors. Ethical and Effective Implementation: Throughout the AI lifecycle, human involvement remains paramount to ensure ethical and effective AI implementation.

The intersection of AI and trust in the political sphere was starkly illustrated in the UK election. AI Steve, the artificial intelligence candidate running for Parliament in the UK's Brighton Pavilion constituency, failed to win a seat in the 2024 general election. This event demonstrates that while AI has made significant strides, there remains a strong preference for human judgment and relatability in areas traditionally dominated by human decision-making.

There's a long way to go from 'In God we Trust' to 'In AI We Trust'.