3x3 Institute

AI Human Trust Relationship

October 2, 2023

Should you (be expected to) trust your AI?

Relationships can be characterized by their degree of honesty, trustworthiness, and a duty to act in the best interest of another party.

Before discussing the user-AI relationship I thought it useful to review relationships to better understand and evaluate the nature of relationships. Many of the aspects of other relationships are likely to be present in the user-AI relationship.

We come across trust relationships all the time:

Some relationships are high trust, some are low trust, and some are in between.

High-trust relationships

High trust relationships emphasize honesty, trustworthiness, and a duty to act in the best interest of another

Example high-trust relationships

What are some example high trust relationships we may have?

Parent-Child: While not strictly a legal or professional relationship, parents are naturally expected to act in the best interests of their children, ensuring their safety, health, and well-being.

Guardian-Ward: In this relationship, the guardian is responsible for the care and well-being of the ward (often a minor or an individual who cannot take care of themselves). The guardian is expected to act in the ward’s best interests at all times.

Doctor-Patient: Doctors are expected to act in the best interest of their patients, providing accurate diagnoses and effective treatments. This relationship is founded on trust, as patients must believe their doctors are giving them the best care possible.

Attorney-Client: Lawyers have an ethical obligation to act in the best interest of their clients. This includes keeping client communications confidential and representing them to the best of their ability.

Investor-Fiduciary: This is a legal or ethical relationship of trust between two or more parties. Typically, a fiduciary prudently takes care of money or other assets for another person. They are legally bound to act in the best interests of their client or beneficiary, even if doing so might be against their own interests. Examples include financial advisors, trustees, and agents with power of attorney.

Agent-Principal: This is a relationship where the agent acts on behalf of the principal. The agent is expected to act in the best interests of the principal, representing them accurately and honestly.

Nature of these relationsips

As I mentioned high-trust relationships are based on trust and the responsibility to act in the best interest of another. These responsibilities often come with legal or ethical obligations attached. There are some variations between them as relationships are not universal and have some variations and limits of duty and scope.

Relationship Doctor-Patient Attorney-Client Investor-Fiduciary
Nature Medical and health-based. Legal and advisory. Primarily financial or asset management but can extend to other areas where trust and advice are paramount.
Duty Doctors are ethically and professionally obligated to provide accurate diagnoses, effective treatments, and to act in the best health interests of their patients. Lawyers have both a legal and ethical obligation to represent their clients zealously and to act in their best interests. Fiduciaries have a legal obligation to act in the best interest of their clients or beneficiaries, prioritizing their clients’ interests above their own.
Scope This relationship focuses on medical care, advice, and health decisions. It also includes the confidentiality of the patient’s medical records and personal information. This relationship focuses on legal advice, representation in legal matters, and decisions related to the law. It includes strict confidentiality obligations regarding client communications and information. This relationship mainly revolves around decisions concerning money, assets, or legal rights. A breach of this duty can lead to legal consequences.
Conflict of Interest Doctors must avoid situations where personal interests could compromise patient care. They must also ensure that they do not have undeclared financial interests in recommended treatments or medications. Lawyers must avoid conflicts of interest. They must not represent clients with opposing interests unless there’s informed consent from all parties. They also need to ensure that personal or financial interests do not compromise their professional judgment and client representation. Fiduciaries must avoid conflicts of interest or, if they are unavoidable, must disclose and manage them transparently.
Examples All healthcare professionals, including general practitioners, specialists, and therapists, engage in this relationship with their patients. All practicing attorneys, regardless of their specialization (criminal law, family law, corporate law, etc.), engage in this relationship with their clients. Financial advisors, trustees, and agents with power of attorney are common examples.

Failure of high-trust relationships

High trust relationships are built on mutual respect, reliability, and consistent positive interactions. They are characterized by open communication, a belief that both parties will act in the other’s best interest, and a history of proven reliability.

Examples:

Parent-Child: Typically, children trust their parents to care for and protect them, while parents trust their children as they demonstrate responsibility. Close Friends: Built over time through shared experiences, these relationships have deep mutual understanding. Long-Term Business Partners: Established through consistent collaboration and mutual benefit. Doctor-Patient: Patients trust doctors to provide accurate diagnoses and effective treatments. Fiduciary Relationships: Such as between a financial advisor and a client, where the advisor is legally bound to act in the client’s best interest. Marriage/Long-term Partnerships: These intimate relationships are generally based on deep trust and shared life goals.

Characteristics:

How High Trust Relationships May Fail:

In conclusion, while high trust relationships offer deep mutual understanding and collaboration, they require ongoing effort to maintain. They can be fragile, and once trust is broken, it can be challenging, though not impossible, to rebuild.

Employee-employer relationship

While not always legally considered a fiduciary relationship, the employee-employer relationship is built on trust. An employee is expected to act in the best interests of their employer, ensuring their work is honest and of high quality. There can be consequences for actions or decisions that harm the company or its stakeholders.

This seemed like a good place to explore a possible model for the user-AI relationship.

Nature of Relationship

The employee-employer relationship is multifaceted and encompasses various aspects, both from a legal and interpersonal standpoint. Here’s a deeper dive into this relationship:

Duties and Responsibilities:

Loyalty and Trust:

Compensation and Benefits:

Rights and Protections:

Performance and Feedback:

Termination:

Personal Growth and Development:

Cultural and Interpersonal Aspects:

In essence, the employee-employer relationship is a combination of legal obligations, mutual expectations, and interpersonal dynamics. Both parties have rights and responsibilities, and the strength of the relationship often determines the success and productivity of the workplace.

Must and employee act in the best interest of the employer?

The expectation that an employee acts in the company’s best interest is a complex issue and can depend on various factors, including the specific employment contract, the nature of the job, and local labor laws. Here’s a nuanced look at the topic:

Duty of Loyalty: Employees generally owe a duty of loyalty to their employers. This means that during their employment, they shouldn’t engage in activities that are directly adverse to the interests of the company. For example, they shouldn’t divert business opportunities away from the company or help a direct competitor.

Conflicts of Interest: Employees should avoid situations where their personal interests might conflict with their duties to the company. If such conflicts arise, they are often expected to disclose them to their employer.

Limits of Loyalty: While employees have a duty to act in the company’s best interest, this doesn’t mean they’re expected to harm themselves, act unlawfully, or go against their core values.

For instance:

Whistleblowing: In many jurisdictions, there are legal protections for employees who report illegal or unethical activities within their company, even if reporting such activities might be against the company’s immediate interests.

Ethical Considerations: Some professions have ethical guidelines that may sometimes conflict with an employer’s interests. For example, a journalist might prioritize the public’s right to know over a media company’s relationship with an advertiser.

Employment Contracts: Some contracts might have specific clauses detailing the expectations for employees regarding company loyalty, non-competition, or handling of potential conflicts of interest.

Local Labor Laws: The specifics of an employee’s duty to their employer can be influenced by local labor laws, which might provide additional protections to workers.

In summary, while employees generally have a duty to act in their employer’s best interest, this duty has its limits, especially when it comes to legality, personal safety, ethics, and individual rights. It’s essential for employees to be aware of their rights and the specifics of their employment contracts. If conflicts arise, seeking legal counsel can be beneficial.

Medium trust relationships

Medium trust relationships sit between high trust and low trust, often characterized by a mix of confidence and skepticism. In these relationships, parties generally trust each other but still exercise caution, often due to past experiences, inherent risks, or lack of complete transparency. Here are some examples:

Peer-to-Peer Online Marketplaces: Platforms like eBay or Craigslist, where users might trust the platform’s basic mechanisms but still approach individual transactions with caution due to potential scams or misrepresentations.

Casual Acquaintances: People you know but don’t share close ties with. You might trust them in casual interactions but might be hesitant to share personal details or secrets.

Coworkers: You might trust them in a professional setting to get their work done, but might be wary about sharing personal grievances or sensitive information.

Franchisee-Franchisor: Both parties might trust each other to a certain extent. The franchisor trusts the franchisee to uphold brand standards, while the franchisee trusts the franchisor for support. However, each may have reservations based on corporate decisions or individual franchise performance.

Banking and Financial Institutions: Customers generally trust banks to handle their money and personal data, but they might be skeptical about hidden fees, interest rates, or the motivations behind certain financial products.

Independent News Sources: Not mainstream, these sources might be trusted for providing alternate viewpoints, but readers might still verify information from additional sources due to potential bias or lack of thorough vetting.

Consultants: While their expertise is recognized, clients might still cross-check some of the information provided by consultants or be wary about the applicability of certain generic solutions to their specific situation.

Online Reviews: While they offer valuable insights, many consumers approach online reviews with a degree of skepticism, aware that some might be fabricated, biased, or based on isolated incidents.

Medical Second Opinions: Trust in the initial diagnosis is not entirely absent, but there’s enough uncertainty or concern to seek an additional perspective.

Service Subscriptions (like streaming platforms): Users trust the platform for content delivery but might be skeptical about price hikes, content removals, or data handling practices.

These medium trust relationships underscore the importance of ongoing communication, transparency, and consistent actions to elevate trust levels. They are often dynamic, with trust levels fluctuating based on experiences and interactions.

Newspaper-Reader

The relationship between a newspaper that accepts advertising and its readers is fundamentally rooted in trust, with the expectation that the newspaper will provide accurate, unbiased, and comprehensive news coverage. Here’s a deeper exploration of this trust relationship:

Expectation of Impartiality: Readers trust that the editorial content of a newspaper is not unduly influenced by its advertisers. They expect news stories to be based on factual reporting and not shaped by commercial interests.

Transparency: It’s crucial for newspapers to clearly distinguish between editorial content and advertisements. This ensures readers can easily identify what is news and what is promotional content. Features like “advertorials” should be clearly labeled as such.

Ethical Standards: Reputable newspapers adhere to ethical standards that separate the newsroom’s operations from the advertising department. This “wall” ensures that advertisers cannot influence news coverage. Journalistic integrity and credibility are paramount to maintaining readers’ trust.

Economic Reality: Readers often understand that advertising revenue supports the newspaper’s operations, allowing them to receive news often at a lower cost or even for free in the case of some online publications. This economic reality can be accepted as long as the newspaper maintains its editorial integrity.

Potential Conflicts: There can be situations where a major advertiser is involved in a controversial or newsworthy event. In such cases, readers trust that the newspaper will cover the story with the same depth, rigor, and impartiality as it would any other story, without shielding the advertiser.

Public Accountability: If a newspaper is perceived to be compromising its editorial content due to advertiser influence, it risks public backlash and loss of credibility. Maintaining trust is not just an ethical imperative but also a business one; a newspaper’s reputation and readership can decline if trust is eroded.

Feedback Mechanisms: Many newspapers have an ombudsman or public editor role, serving as an intermediary between the readers and the newsroom. This role often addresses readers’ concerns about potential biases or influences, ensuring that the trust relationship is maintained.

In summary, the trust relationship between a newspaper taking advertising and its readers is delicate and hinges on the newspaper’s commitment to editorial integrity, transparency, and accountability. While readers often understand the economic necessity of advertising, they expect clear boundaries to ensure unbiased and accurate news reporting.

Consultant-Client

The trust relationship between a consultant and a client is fundamental to the success of their collaboration. This relationship is characterized by a mutual understanding that both parties will act in good faith to achieve a common goal. Here’s a detailed look at this trust relationship:

Expertise and Credibility: Clients trust consultants based on their expertise, experience, and credentials. They expect the consultant to possess the necessary skills and knowledge to address their specific needs.

Confidentiality: A significant aspect of the trust relationship is the consultant’s duty to maintain confidentiality. Clients often share sensitive information, and they trust the consultant to safeguard and not misuse this information.

Objectivity and Honesty: Clients trust that consultants will provide unbiased advice, even if it’s not what the client wants to hear. They expect the consultant to prioritize the client’s best interests over any personal or third-party interests.

Transparency: Clients expect clear communication regarding the consultant’s methods, the progress of the project, any potential conflicts of interest, and the basis for any recommendations. Transparency helps in managing expectations and building trust.

Reliability: Meeting deadlines, delivering on promises, and being responsive to client inquiries are essential for maintaining trust. Reliability assures the client of the consultant’s commitment and professionalism.

Value for Money: Clients trust that the fees they pay are commensurate with the value they receive. Overcharging or underdelivering can quickly erode this trust.

Ethical Standards: Adherence to professional and ethical standards is crucial. This includes not only maintaining confidentiality but also avoiding conflicts of interest and being honest about the consultant’s capabilities and the project’s feasibility.

Feedback and Adaptability: Trust is fostered when consultants actively seek feedback and are willing to adapt their approach based on client needs and feedback.

Open Communication: An open channel for communication allows clients to voice concerns, ask questions, and stay informed, all of which reinforce trust.

Long-Term Relationship: Many consulting engagements aim for long-term collaboration. A consistent record of success, integrity, and mutual respect deepens the trust over time, often leading to repeat engagements or referrals.

In summary, the trust relationship between a consultant and client is multi-dimensional, anchored in expertise, transparency, ethical behavior, and consistent delivery of value. This trust is both fragile and invaluable; once broken, it can be challenging to restore, but when maintained, it forms the foundation of a productive and long-lasting partnership.

Web search engine-User

The trust relationship between a web search engine and its users is crucial for the search engine’s success and reputation. As with other relationships, it’s rooted in mutual expectations and understandings. Here are some facets of this trust relationship:

Relevancy of Results: Users trust a search engine to provide accurate and relevant search results. They believe the engine’s algorithms will return results that are closely aligned with their queries.

Privacy Concerns: Trust is established when users believe that their search queries, personal data, and browsing habits are treated with respect and confidentiality. Any breaches or perceived misuse of this data can erode trust significantly.

Transparency: Users expect transparency in how search results are generated and presented. For instance, they’d like to know which results are organic and which ones are paid advertisements.

Unbiased Information: While it’s understood that search engines might prioritize paid content, users trust that the algorithms, by and large, aren’t biased by external pressures or the search engine’s own interests. They expect a certain level of neutrality.

Safety and Security: Users trust search engines to flag or filter out malicious sites or content that could harm their devices or steal their data. A secure and safe browsing experience is foundational to this trust.

Continuous Improvement: Search engines are expected to continually refine and improve their algorithms to provide better, more relevant results over time.

User Experience: A clean, user-friendly interface that’s free from excessive ads and is easy to navigate fosters trust. An overcrowded or confusing interface can detract from the user’s trust in the search engine’s effectiveness.

Feedback Mechanisms: Providing users with avenues to give feedback or report issues, and showing responsiveness to such feedback, can enhance trust.

Adherence to Regulations: Compliance with data protection regulations, such as the General Data Protection Regulation (GDPR) in the European Union, plays a crucial role in establishing trust, as it ensures that user data is handled appropriately.

Cultural and Regional Sensitivity: Recognizing regional nuances, languages, and cultural differences in search results can foster trust among users from various backgrounds.

In summary, the trust relationship between a web search engine and its users is shaped by the search engine’s ability to deliver relevant results, ensure user privacy, maintain transparency, and offer a safe browsing experience. As with other trust relationships, once compromised, it can be difficult for a search engine to regain user trust.

Low trust relationships

Low trust relationships are characterized by skepticism, doubt, and the expectation that one party may not act in the best interest of the other. Here are some examples of relationships that can often be characterized by low trust:

Buyer-Seller in Used Goods Markets: Especially in scenarios where there’s no warranty or return policy, buyers might be skeptical about the quality and authenticity of the product being sold.

Landlord-Tenant: This relationship can sometimes be fraught with mistrust, with tenants doubting the landlord’s responsiveness to issues or the return of their deposit, and landlords wary of potential property damage or unpaid rent.

Competing Businesses: Companies in direct competition often view each other with mistrust, especially in industries where corporate espionage or unethical business practices might be prevalent.

Politician and Public: In many regions, politicians are viewed with skepticism by the public due to perceptions of broken promises, corruption, or the influence of special interest groups.

Law Enforcement and Certain Communities: In areas where there have been instances of police misconduct or discrimination, the relationship between law enforcement and the community can be marked by deep mistrust.

Employee-Employer: In companies with a history of layoffs, unfair practices, or poor communication, employees might not trust the company’s leadership or future promises.

Online Scams and General Public: Unsolicited communication from unknown sources, especially those asking for personal details or money, are typically met with mistrust.

Certain Media and the Public: Media outlets that have been accused of bias, misinformation, or sensationalism might be viewed with skepticism by certain sections of the public.

Neighbors in Dispute: Neighbors who have had disagreements or conflicts over property boundaries, noise, pets, or other issues might view each other with mistrust.

Contractors and Clients: Especially in the absence of contracts or clear communication, clients might be wary of potential overcharging, delayed work, or subpar service quality from contractors.

It’s important to note that while the relationships listed above can sometimes be characterized by low trust, they aren’t universally so. Trust is dynamic and can be built or eroded based on individual experiences, actions, communication, and efforts to rectify past issues.

Ok then how about our AI relationships?

The trust relationship between an AI system and its users is especially critical, given the increasing reliance on AI for decision-making and various tasks. This trust relationship is influenced by several factors:

Accuracy and Reliability: Users need to trust that the AI will perform consistently and provide accurate results. If the AI system makes frequent errors or is unpredictable, trust is eroded.

Transparency and Explainability: Understanding how an AI system makes decisions can bolster trust. If the system’s logic is opaque or too complex to comprehend (often referred to as a “black box”), users might be more hesitant to trust its outputs.

Data Privacy: Users must trust that any personal or sensitive data input into the AI system is handled securely and won’t be misused or shared without appropriate consent.

Bias and Fairness: Trust is influenced by the perception of fairness. If an AI system is found to be biased or discriminatory (e.g., in areas like hiring, lending, or law enforcement), trust in the system and its developers can be severely compromised.

Robustness and Security: Users need to believe that the AI system is resistant to adversarial attacks and won’t malfunction when confronted with unexpected inputs.

Feedback and Adaptability: AIs that learn from user feedback and adapt over time can increase user trust. Conversely, systems that don’t evolve or adapt to user needs might be seen as less trustworthy.

Clear Boundaries: Users should be aware of the AI’s limitations. Overselling or misrepresenting an AI system’s capabilities can lead to misplaced trust.

Human-AI Collaboration: Trust can be enhanced when AI systems are designed to work alongside humans, complementing human abilities rather than replacing them. For example, an AI that suggests options and explains its reasoning, while allowing the human user to make the final decision, can foster trust.

Ethical Considerations: Users want to believe that the AI system operates under ethical principles. This includes not only avoiding bias but also ensuring the system is used for beneficial and not malicious purposes.

Regulations and Standards: Compliance with established standards and regulations can enhance trust, as it assures users that the AI system meets certain benchmarks for safety, fairness, and reliability.

Accountability: Knowing that there’s a clear line of responsibility (be it an organization, developer, or regulatory body) in case the AI system fails or causes harm can influence user trust.

In summary, trust between an AI and a user is multifaceted and shaped by the AI’s performance, transparency, ethical considerations, and adaptability, among other factors. As AI systems become more integrated into daily life and decision-making processes, establishing and maintaining this trust will be increasingly vital.

Can an AI user relationship be high trust?

Yes, an AI-user relationship can be high trust, but the dynamics differ from traditional human-human relationships. The foundation of trust in an AI-user relationship primarily rests on the consistent performance, reliability, and perceived safety of the AI system. Here’s how an AI-user relationship can achieve high trust:

Consistent Performance: If the AI consistently delivers accurate and reliable results, users will place more trust in it.

Transparency: Users are more likely to trust AI systems when they understand how they work. Transparency in decision-making processes, especially in explainable AI models, can enhance trust.

Privacy and Security: Ensuring that user data is handled securely and responsibly is paramount. Users must trust that their interactions with the AI will not compromise their privacy.

User Experience: A user-friendly, intuitive interface that’s free of glitches and offers a smooth experience can bolster trust.

Feedback Loops: AI systems that incorporate user feedback and adapt or improve based on it can help in building trust.

Ethical Behavior: AI that is perceived to operate within ethical bounds, avoiding biases, and ensuring fairness, will earn greater user trust.

Predictability: Users should be able to anticipate how the AI will behave based on their inputs. Unexpected or erratic behavior can erode trust.

Clear Boundaries: Users need to understand an AI’s limitations. Clearly communicating what the AI can and cannot do will prevent misplaced trust.

Accountability: If something goes wrong, users need to know that there’s a clear line of responsibility, whether it’s the developers, operators, or another entity. This accountability can instill more trust in the system.

While it’s possible for an AI-user relationship to be characterized by high trust, it’s essential to remember that “trust” in this context doesn’t carry emotional reciprocity as in human relationships. Users trust the system’s functionality and reliability rather than forming an emotional bond. Additionally, maintaining this trust requires ongoing diligence, as lapses in any of the above areas can quickly erode the trust users have in an AI system.

What about competence and trust relationships?

We know that AI performance isn’t always what we expect. How does competence fit in?

Competence is a critical component of trust in many relationships. When individuals believe that another person is capable and knowledgeable in a specific domain, they are more likely to trust them. Here’s how competence relates to trust relationships:

Foundation of Professional Trust: In professional settings, competence is the bedrock of trust. Whether it’s doctors, engineers, pilots, or any other professionals, their expertise and ability to perform their job effectively is crucial for trust. Patients trust doctors because they believe in their medical competence, and passengers trust pilots because they trust their flying skills.

Meeting Expectations: Competence ensures that expectations are met consistently. When someone consistently delivers results or meets expectations, trust in their abilities solidifies.

Predictability: Trust often grows from predictability. When someone is competent, their actions and results are often consistent and predictable, further enhancing trust.

Safety and Reliability: In many situations, trust in someone’s competence equates to trust in one’s safety. For instance, trusting the competence of a car mechanic can mean trusting that the car won’t break down unexpectedly.

Mitigating Uncertainty: Trust in competence can reduce the anxiety of uncertainty. For example, even if someone doesn’t understand the intricacies of a legal issue, they trust a competent lawyer to navigate it for them.

Enhancing Communication: When individuals believe they’re dealing with someone competent, they’re more likely to engage in open and productive communication, believing that the exchange will be fruitful.

Failure and Trust Erosion: While consistent competence builds trust, repeated failures or mistakes can rapidly erode it. Once someone’s competence is in doubt, restoring that trust can be challenging.

Overconfidence: There’s a difference between competence and overconfidence. Someone who is overconfident might take unnecessary risks, which can jeopardize trust.

Interpersonal Relationships: While competence in skills or knowledge is vital in professional settings, in personal relationships, competence might relate more to emotional intelligence, understanding, and the ability to manage relationship dynamics effectively.

Growth and Adaptability: Part of perceived competence is the ability to learn, adapt, and grow. In a rapidly changing world, trust is often placed in those who show they can evolve and handle new challenges.

In conclusion, competence is a foundational element of trust in many relationships. Whether it’s trusting someone’s skills, knowledge, or their ability to handle interpersonal dynamics, competence plays a pivotal role in establishing and maintaining trust.

Concluding thoughts

One aspect that has not been discussed so far is the concept of “Vulnerability” in trust relationships:

Trust inherently involves vulnerability. When you trust someone, you’re essentially placing yourself at some degree of risk, whether emotionally, physically, or otherwise, believing that the other party won’t exploit or harm you. Here’s a breakdown:

Emotional Vulnerability: This is most evident in personal relationships. When you confide in someone, you’re trusting them with your emotions and secrets. Betrayal of this trust can lead to emotional pain.

Financial Vulnerability: In business or monetary dealings, trusting someone with investments or business decisions means you’re at risk of financial loss if they don’t act in good faith.

Physical Vulnerability: This can be seen in scenarios where you trust someone with your safety. For example, trusting a driver when you’re a passenger in their vehicle or trusting a surgeon with a medical procedure.

Reputational Vulnerability: Trusting someone with sensitive information might put your reputation at stake. If they misuse the information, it could lead to personal or professional embarrassment.

Dependency and Power Dynamics: Trust often means relying on someone, which introduces power dynamics. Those with more power or control need to be especially cautious not to misuse this trust.

Trust as a Leap of Faith: In many situations, there isn’t enough information available to guarantee an outcome. Trusting in such scenarios is a leap of faith, accepting the vulnerability that comes with uncertainty.

Healing and Rebuilding: Post-betrayal or after trust has been broken, recognizing and addressing vulnerabilities becomes paramount. Rebuilding trust requires acknowledging the areas of vulnerability and working to ensure they’re not exposed to potential harm again.

In essence, trust relationships always entail some form of vulnerability. Recognizing and understanding this vulnerability is crucial for nurturing and protecting trust. It emphasizes the responsibility that comes with being trusted and the care that should be taken not to exploit or damage that trust.