Claiming status as a professional means owning and aspiring to achieve the standards for the engagements we have with others. The codes that govern professional roles all derive from five principles, which need to be closely reviewed in relation to the use of AI coaching tools. So, for each principle, we need to explore some basic questions to avoid the hazards posed by the AI phenomenon.
Standard 1: Fair Play
Are the rules clear, and are resources equally available to the learner and the service provider? When it comes to the use of AI, do the learners involved know how the AI engine is going to use their inquiries and responses? Do the learners have a clear sense of control over their content and how that content will be protected? How can a learner dispute a finding? Are the “rights” of the learner easily available?
Standard 2: Autonomy
Individuals need to be free to opt in and out as their needs change and need to feel their interactions are safe. How does the AI coach engage with the learner—through voice, writing, or visual channels? What information is kept and is shared? What choices do the learners have in managing the engagement?
Standard 3: Fidelity to Evidence
Information provided should be clearly delineated in terms of sources, provenance, and limitations. Does the AI coach use an open source, scraping the web to produce answers? Does the AI coach use a curated, research-based, fully vetted database that is managed for validity and reliability? Can the provider of the AI coach identify the information sources used by the engine, and if questioned, outline specifically the nature of and limits of the database being used? If it is open source, does the learner know that AI Coaches lie, misrepresent, and hallucinate? That they are programmed with a bias to answer confidently, even if incorrectly.
Standard 4: Beneficent Intentions and Outcomes
Does the AI coach have a persona designed to be in a coaching voice? Does the programming related to the AI coach provide constructive answers, avoiding bias, harmful assertions, and distorted comments based on a range of stereotypes? Is the AI coach geared toward constructive actions related to individual growth or toward dealing with others?
Standard 5: Non-maleficence
The core of this standard is knowing and addressing the potential of unintended consequences. LLMs have been proven to misrepresent, create false narratives, hallucinate, and, no matter what the companies claim about the efforts to reduce these flaws, they continue to be present. Multiple tests of supposedly highly filtered AI coaches that use web-based scraping cannot escape the distortions and misguided answers that are sometimes provided. Does the AI coach you plan to use for your learners have the protections that only a vetted and curated database can provide, and does it not scrape the open web for data?
Professional Standards
Professional standards are intended to provide protection for both the professional and the learner—or client. The provision of caring, thoughtful, and proven guidance is a minimum requirement. Already, there are legal cases addressing all of the standards above, which means professionals using open platforms to provide services are merely standing in the middle of the road asking to be run over. The legal and moral risks may seem distant or even illusory, but these are real. Who wants to carry the burden of having arranged for the use of a service that does harm to others? Who wants to stand in front of the CEO who asks, “Who approved the use of this AI coach, which has encouraged our top performers to leave? Or worse, an AI coach whose advice results in injury or death?
The bells and whistles of the tsunami of AI coaches that are being provided are the contemporary Sirens of ancient myth, luring others to unhappy conditions with their enchanting features. How many will crash on the shores before we wake up to ask why we ignored the guidance our professional standards have provided, and our eyes and ears are regularly taking in the news of the limitations of this new technology as it relates to human well-being. Some solutions avoid these issues, and you have to demand them to avoid the pitfalls outlined above.
