Topic from the field: “Should artificial intelligence replace the addiction counselor?”

I’ve recently been presented with the topic of artificial intelligence (AI) replacing addiction counselors.  This post constitutes the current condition of my thinking about that question.

By the way, I’ll say that artificial intelligence is an interesting topic on its own.  And on its own it’s one I’m not prepared to comment on in a competent way. 

Rather than ask, “Will AI replace the addiction counselor?”, I have designed the question as “Should?”  This is because I’d rather avoid prediction as the topic than open the door to a possibility.  And this work will explore the possibility.

I’ll say to the reader that as I wrote that sentence above about “will” vs. “should”, I wondered if AI would make such a consideration – prior to or while composing.  I’m concerned it would not.  And so, some of the features inside the human compositional process of spoken or written language that can or could make that consideration are closer to the topic of this work.

The master soup maker

The now late Lee Feldman spent many hours with me over a series of years, several years ago.  I have shared information and challenges Lee presented to me in previous works here on Recovery Review.

One that I’ll share now relates to the word “qualia” and Lee’s notion of “the master soup-maker”.

Lee introduced me to the word “qualia”:  the subjective aspects of human consciousness¹.  To help deliver this idea to me he used the example of eating a peach.  He told me how one who has eaten a peach cannot in any words transfer what it is like to have the experience of eating a peach to someone who has not.  Or even to someone who has. As Lee put it, qualia cannot be transferred from one person to another.

He then expounded his work on the first computerized psychotherapist.  He made a wide range of fascinating points that are far too many and lengthy to share here. 

During that discussion, Lee described to me the development of a complex multi-factorial decision making heuristic for automated production of high-quality soup with a consistent flavor.  Lee told me how the attempt to automate this in a factory with industrial-sized batches never succeeded in matching the acumen of “the master soup maker” within the factory.  And that ultimately the “solution” was to “continue to apprentice master soup makers”, rather than hand this work off to a machine.

In doing so, Lee expounded how the qualia of the master soup maker were the sole source of valid guidance in adjusting the real world application of the recipe and cooking method to variables that could not be controlled.  For example, Lee explained that the master soup maker could not prevent having to use differing flavors of milk (that differ in flavor based on the time of year the cow was fed and milked) in different batches of the same soup, and adjusting the recipe accordingly upon tasting the “final” product before finishing.

And so, I ask the reader if they would prefer a counselor that:

• ignores the real conditions of patient qualia;
• ignores the qualia within the “third” of the therapeutic alliance;
• and merely perseveres in the application of a formula. 

Introspection and Empathy

In his article titled “Introspection, Empathy, and Psychoanalysis:  An Examination of the Relationship between Mode of Observation and Theory”² Heinz Kohut expounds the capacity for introspection as an essential ingredient in psychoanalysis.  In doing so he provides two details. 

• One is that introspection is the literal observational tool of the patient and also of the analyst.  He compares introspection to a telescope for an astrophysicist or a microscope for a biologist.  He describes both the capacity for introspection and the function of introspection.  It’s a fascinating read. 
• The other detail is the location of the introspective observational method:  a person’s own psychological interior.  He notes what the content in that location is and how it cannot be observed by any method other than introspection.

Kohut continues by expounding how the patient introspects their own interior, while the analyst introspects the analyst’s own interior. 

But he goes on to say that the analyst also participates in an additional function, which is the knowledge-based expert observation (both seeing and listening) of the patient’s exterior, and forming judgments about the material accessible to (observable by) the analyst.  And in doing so he provides two details.

• One is to differentiate subjective observation (introspection) from objective observation (listening and watching).
• The other is to differentiate “experience-near” methods from “experience-distant” methods. 
  • A rough example of an experience-distant connection would be a clinician who understands the patient based on the clinician’s personal life experience or from seeing many other patients.  In our field a recovering counselor, recovery coach, or peer support professional would at best understand from the “experience-distant” perspective, in spite of them being able to “relate”.
  • A rough example of an experience-near connection would be a clinician whose defenses are lowered, who has surrendered their goals for the patient, and does not have intrusive memories of this patient’s previous sessions in the moment.  And who is in that position or attitude while prioritizing attunement, plus nothing. 

Why does Kohut describe the analyst introspecting their own interior? 

One reason is the idea that the analyst’s own psychological experiences (thinking, memories, affects, feelings, defense functions, etc.) that seem to emerge during sessions are themselves considered to be important content – imparted or evoked by the patient. 

• That is to say, countertransference in a simple sense is at least an entire second domain of information about the material of the patient (where the first domain is simply what the patient says or otherwise displays). 
• Put more simply, material evoked in the analyst by the patient is material about the patient.

By a rough approximation on my part all of these, then, are the variety of raw ingredients that comprise what Kohut understands empathy to consist of.  And in that paper Kohut outlines how introspection and empathy in combination are a more therapeutically effective factor than the expert knowledge delivered with expert technique by the analyst. 

Yes, in that way Kohut put “empathy and introspection” above “truth” on a potency hierarchy of active ingredients. 

In this he broke from Freud who put forward the notion that truth was the most potent active ingredient in the psychoanalytic arrangement, and that the technique of delivering interpretations to convey truth was the vital skill of the clinician. 

And so, I ask the reader, at this point in the developmental history of AI:

• Does AI have the empathic and introspective capacities Kohut describes – such that AI matches a working clinician in the effectiveness of those capacities? 
• Or does AI only deliver truth given its inherent limitations?  And if so, is that sufficient?
• Are we satisfied to use an AI that accumulates facts and delivers information (truth), or do we agree with Kohut that empathy is more potent and required, while finding it missing from AI?

Technique vs Empathy

One puzzle of providing addiction counseling per se is figuring out how to respond to the presenting level of “ready-willing-able”, with respect to person-centered goals and clinical engagement of the patient. 

In a post titled “Technique vs. Empathy”³  I outlined three potential options in a therapeutic situation:

1. The counselor’s technique is correct and the patient must adapt to it.
2. The counselor must adapt to the technique of the patient.
3. Empathize, and neither lead nor follow the patient.

In doing so I described how each of these can stand alone, how each combination of any two of them can function, and how all three combined can help us see our duty for attunement. 

• That is to say, the clinician’s capacity to determine which path to adopt, and to oscillate between these options (accurate flexibility) is key, rather than being stuck (such as staying within one method, or abruptly flipping between pre-packaged methods, etc.).
• Further, innovating in the moment would rely upon and allow novel blending of these ingredients.

And so, I ask the reader, does AI have the meta-cognition required to blend or oscillate between these ingredients as a function seated in clinical creativity?

What is empathy?

The abolition of asking questions can be one aspect of trauma-informed care.  The reason for a clinical capacity for a question-free approach is the recognition that being asked a question can be experienced as aversive.  And that a whole series of questions can be more aversive. 

It has been said that addiction counseling often seems to over-rely on the asking of questions.  That is to say, the historical judo-like cajoling of the addiction treatment patient into compliance with the counselor’s method and goal relies on the one technique of asking a series of questions.  Questions that are meant to produce change. 

In a work titled Is it too late to ask, “What Is Empathy?”⁴ I wrestled with the topic of what empathy is.  And I did so in such a way that the title and every sentence of the entire work were nothing but questions.  My hope was that it would give the reader an experience of how poorly a series of questions can be experienced – that a series of questions is not necessarily a recipe to convey empathy.  Aside from the content as questions, the work might be worth a read if you’d like to struggle along with me on the topic of what empathy is.

And so, I ask the reader, does the current state of AI suffice in a trauma-informed way?  Does AI know the best timing for the use and the non-use of questions?

AI will not weep while you cry

During one’s counseling one might cry.  And the therapist might weep a bit while the patient cries.  I believe I barely need to say anything at all to expound that point.  I’ll just say that sometimes being mirrored by someone is evocative of something deep in our human experience.  And at times that experience provides both relief and something that is sustaining.  In that context, two written works come to mind.^{5, 6}

Idiom and the unconscious

To help convey the device of idiom, the area of the unconscious, and the relevance of these to the advisability of AI replacing addiction counselors, consider 3 different scenarios.

Scenario 1:

• Patient:  “Every time my boss talks with me, I see red.”
• Therapist:  “Tell me all the ways your boss is important to you.”

I wonder if AI understands the idiomatic content of various languages (e.g. culture-bound connotation vs. explicit denotation) accurately.  That is, I wonder if AI is culturally competent in at least the domain of language.    

Scenario 2: 

• Therapist:  “Tell me about this painting.”
• Patient:  “I don’t know what that’s a painting of, but it’s not a painting of my mother!  She’s the furthest thing from my mind!”

I wonder if the AI counselor has the capacity to function accurately within the unconscious content of language, and to render an understanding accordingly.  Frankly, novice counselors generally do, but require apprenticing to make good use of this capacity.

Scenario 3:

• Patient:  “Throughout my whole life, it seems like everyone abandons me.”
• Therapist:  “So now you manufacture your own abandonment.  Fascinating.  I wonder why.  If I knew, it would probably make sense.”

I wonder if AI has the capacity to consider and bear the possibility of a domain of information that is present only in a disguised version, and that is out of the conscious awareness of both the patient and the therapist.  And to incorporate that domain in the work when it emerges.

One layer of content in therapy is the layer comprised of the conscious, the present, and the willed.  And another layer is that comprised of the unconscious, the past, and the wished.^{7, 8}  Regardless, all are communicated through language – and therein lies the difficulty.^{9, 10, 11} 

Andrea DeSantis, Unsplash

Patient preference vs effectiveness

In my conversations with Lee Feldman, I told him how the CEO at my previous workplace, the now late Mike Boyle, was fond of following the area of AI counselors and was convinced computers would someday be far better counselors than any human could ever be.  Mike remarked how in one study after patients had been randomly assigned to AI counselors or human counselors, they were informed to which they had been assigned, offered the chance to remain in the counseling as a thanks for participation, and the opportunity to pick an AI or a human counselor regardless of which they had during the study.  Mike shared that the group with an AI counselor overwhelmingly chose to stay with AI.  The researchers were amazed and asked why.  In sum, the patient feedback was that they preferred the AI counselor to a human counselor, as an AI counselor didn’t present what Mike called “nutty therapist variables” like talking about their weekend as an introductory warm-up for the session.

While that might seem compelling, I ask the reader if those selecting AI in that study might have been unconsciously avoiding or pursuing something.

And I ask the reader to consider the research literature^{12, 13} describing how patients prefer physicians in clinical settings to continue to wear white lab coats – even though doing so is not best practice, as lab coats contribute to the spreading of pathogens. 

As Norm Hoffmann has put it, “Do we want to ask the patient what they prefer, or what they found helpful?”

The value proposition of silence

This entire work could be reduced to reducibility and silence, as follows.

1. Reducibility.¹⁴  Is the following true or false? 
   • “What is found in psychology is actually reducible to biology.  And what is found in biology is actually reducible to chemistry.  And what is found in chemistry is actually reducible to physics.  And what is found in physics is actually reducible to math.”  My answer is “false”.
   • And thus, AI at this time can not do the job of the addiction counselor.
2. The value proposition of silence.
   • Who is to determine this and on what grounds? 
   • And thus, AI at this time can not do the job of the addiction counselor. 

---

References

¹ “In philosophy of mind qualia (/ˈkwɑːliə, ˈkweɪ-/; sg.: quale /-li/) are defined as instances of subjective, conscious experience.  The term qualia derives from the Latin neuter plural form (qualia) of the Latin adjective quāils (Latin pronunciation: [ˈkʷaːlɪs]) meaning “of what sort” or “of what kind” in relation to a specific instance, such as “what it is like to taste a specific apple — this particular apple now”. 

Examples of qualia include the perceived sensation of pain of a headache, the taste of wine, and the redness of an evening sky. As qualitative characteristics of sensation, qualia stand in contrast to propositional attitudes, where the focus is on beliefs about experience rather than what it is directly like to be experiencing.” (Wikipedia.  Retrieved on 09/01/2024).

² Kohut, H.  (1959).  Introspection, Empathy, and Psychoanalysis:  An Examination of the Relationship between Mode of Observation and Theory.  Journal of the American Psychoanalytic Association. 7(3): 459-483.

³ Technique vs. Empathy (recoveryreview.blog)

⁴ Is It Too Late to Ask, “What Is Empathy?” (recoveryreview.blog)

⁵ Kurtz, E. & Ketcham, K. (1993).  The Spirituality of Imperfection:  Storytelling and the Search for Meaning.  Bantam.

⁶ Love and Addiction Counseling (Bill White and Jason Schwartz) (recoveryreview.blog)

⁷ Amundson, J.  (1981).  Will in the Psychology of Otto Rank:  A transpersonal perspective.  The Journal of Transpersonal Psychology. 13(2): 113-124.

⁸ Wadlington, W.  (2012).  The Art of Living in Otto Rank’s Will Therapy.  The American Journal of Psychoanalysis.  72: 382–396.

⁹ Rescorla is to Pavlov as Semiotics is to Freud (recoveryreview.blog)

¹⁰ Question from the field: What makes an addiction counselor a “Master Craftsman”? (recoveryreview.blog)

¹¹ Addiction:  Understandings and Enactments of the Current Era (recoveryreview.blog)

¹² Haun, N., Hooper-Lane, C. & Safdar, N. (2016).  Healthcare Personnel Attire and Devices as Fomites: A Systematic Review.  Infect Control Hosp Epidemiol.  37:1367–137.

¹³ Petrilli, C.M., Saint, S., Jennings, J.J., Caruso, A., Kuhn, L., Snyder, A., & 2 Vineet Chopra, V.  (2018).  Understanding Patient Preference for Physician Attire: A cross-sectional observational study of 10 academic medical centres in the USA. BMJ Open. doi:10.1136/ bmjopen-2017-021239.

¹⁴ What Biases Do You Observe Among Many of the Scientific and Medical Experts in the Field? (recoveryreview.blog)

---

Extra Credit Readings

Coon, B.  October 17, 2020.  Should We Include a Moral Dimension? The Aesthetics and Anesthetics of Addiction (recoveryreview.blog)

Coon, B.  January 15, 2021.  Negative space (recoveryreview.blog)

Coon, B.  February 20, 2021.  Recovery: What Is It Good For? (recoveryreview.blog)

Coon, B.  February 27, 2021.  “Throw Flour On the Invisible Man”: Toward locating recovery function and assessing recovery quality (recoveryreview.blog)

Luft, J. & Ingham, H. (1955). The Johari Window, a Graphic Model of Interpersonal Awareness. Proceedings of the Western Training Laboratory in Group Development.  Los Angeles: UCLA.

Rowe Jr., C & Mac Issac, D. S.  (2002).  Empathic Attunement: The “Technique” of Psychoanalytic Self Psychology.  Jason Aronson, Inc.

Solms.  M.  (2022).  The Hidden Spring:  A Journey to the Source of Consciousness.  W. W. Norton & Company.

Solms, M., & Friston, K. (2018). How and why consciousness arises: Some considerations from physics and physiology. Journal of Consciousness Studies. 25(5-6), 202–238.

Solms. M. & Turnbull, O.  (2003).  The Brain and the Inner World:  An Introduction of the Neuroscience of Subjective Experience.  Other Press.

---

Suggested Readings

Here are two readings I do suggest, that are accessible and far more digestible than the Extra Credit readings listed above.

The first one is relevant in a direct way. The second one uses a practical problem with a concrete solution as a window into our work.

“The patient’s unknown goal”

Coast Guard Search and Rescue:  Lessons and Inspiration

---

Lee Feldman’s bio

Access from the world wide web on 09/17/2024 at the following address: https://webservices.ncleg.gov/ViewDocSiteFile/71117

Lee T. Feldman:
Integrya, LLC: Co-Chair
 Genii.us: Chief Scientific Officer
 Æsir Management, Inc.: Management Consultant
 Institutes for Human Security: Executive Director, Member of the Managing Board

Current Positions: Co-Chair Integrya, LLC & Chief Scientific Officer for three subsidiary consultancies: Genii.us, Æsir Management, and Executive Director and Member of the Managing Board of the Integrya not-for-profit affiliate – The Institutes for Human Security/National Peace Foundation.

At Integrya and its affiliated consultancies, Lee’s focus is on providing consulting and research related to mapping scientific information into policymaking and legislative activities that are substantially dependent on an accurate foundation of information and quantification of uncertainties. At The Institutes for Human Security/National Peace Foundation, Lee serves as Executive Director, and Member of the Managing Board. His focus is on coordinating a number of subsidiary institutes that address global threshold issues in health, climate, civil society development, and other human security and human ecology arenas.

Although his graduate studies were in immunochemistry and physics, Lee began working with the US Department of Defense in 1973 on novel radar suppression technologies. His science policy work began in 1977 with the Defense Intelligence Agency where he served as Chief Scientist for the U.S. Indications and Warning Intelligence Task Force. This work led to Feldman’s development of the High Integrity Policymaking Methodology [HIP], a process and toolset that helps decision-makers apply quantitative fact-finding methods to integrate the scientific, technical, and ethical aspects of policymaking. A particular emphasis of Lee’s work is on methods for evaluating the integrity of the foundation information used in a policymaking process with a focus on quantifying uncertainty and increasing appreciation of risks that derive from the unexpected and unknown.

Integrya applies the HIP approach to develop private-sector solutions to a broad range of programs in developing global markets, healthcare and care coordination, environmental impact, human ecology, security, and technology. The Institutes for Human Security continues Lee’s basic research on HIP, and applies these same methods to work on global challenges and international policy related to health, human potential, sustainable development, developing normative economic approaches in emerging and less-advantaged economies, health diplomacy, and empowering Citizen Peace-Builders to improve the institutions of civil society.

As a scientific and policy consultant, Lee has applied the HIP methodology to policy issues in healthcare, pharmaceutical and biological technologies, biodefense policy, emerging infectious disease policy, emerging and developing market econometrics, communications technology, defense and intelligence systems, national security policy, banking and finance, international intellectual property and global trade policy, and strategy development related to environmental policy, developing world relations, and global food, nutrition and obesity issues. Major government and multilateral organization engagements include: the US Department of Defense, US Trade Representative, US Department of Health & Human Services, US Department of Commerce, several Canadian federal and provincial health ministries, Namibia, Kenya, Romania, United Kingdom, European Parliament and, at the UN, the World Health Organization, UNAIDS, the World Trade Organization, the World Intellectual Property Organization. Lee has also worked to apply the HIP methods to several international election observer missions (Pakistan, Haiti, and Philippines where HIP models were employed to help identify election fraud and abuse). Major recent commercial engagements include: Sanofi-Aventis, Pfizer, General Electric, Siemens, Medtronic, Microsoft, IBM, and a number of information and biotechnology companies and several investment banks and venture capital firms.

Lee is also on the Clinical Advisory Board of CRC Health Group [the nation’s leading provider of treatment and educational programs for adults and youth who are struggling with behavioral issues, chemical dependency, eating
disorders, obesity, pain management, or learning disabilities], and Afina Advisors [a investment banking firm focused on market opportunities in Latin America]