Psychiatrists

There's a fundamental distinction lacking in these comments between the existence of AI "psychiatrists" and their effectiveness. The fact of the matter is that AI therapists (and diagnosticians) already exist. And, irrespective of whether it's in their best interests, people are already using GP AI in this way. The debate has shifted to the point of whether AI *should* be used and how effective it'll be. I agree with the argument that you can't form the same connection with a machine and that the fundamental social neurochemistry/neurobiology is different. I even agree that parasocial AI relationships are only likely to exacerbate users' problems in the long run. But now you have to convince upcoming generations, conditioned to use AI agents as surrogates for friends and even partners, that this is the case. I think this is the core conundrum that flesh and blood psychiatrists and allied workers (psychologists, psychotherapists, counsellors etc.) are facing down. Classic Jurassic Park reference: it's not whether they can, but whether they should. They definitely can - and, among the younger generations, *are* - but they probably shouldn't. Now it's over to you to persuade them. Yes, psychiatrists are at a distinct advantage as full doctors with prescription abilities. It's going to be quite the hurdle to get regulatory agencies to accept AI agents as the sole point of call. It's less of a burden, however, to get them to accept AI as a "supplement" to daily human practice (which is already happening with or without approval, especially with academic experiments indicating that top-line AI models are outperforming specialists, on average, at diagnoses and handling sequelae across all fields). If you're seeing a young doctor of any stripe today, there's a fair chance they're incorporating AI into their grunt work. From there, it would seem to be a slippery slope from AI transcription to "intelligent" suggestions to asking medical questions. I see psychiatrists being only modestly more isolated from risk than other forms of therapist over a 20 year timescale (which is the true subject of this site), but it's enough to tip me from my usual answer of "Moderate" into "Uncertain", particularly with the burden of accountability surrounding the dispensation of some extremely heavy medications like lamotrigine, clozapine, etc. How is AI, for example, going to deal with the rare but life-threatening SJS caused by lamotrigine? What does that look like? An app on your phone that follows up on high-risk medications? Would it be that different to running to A&E when a medication causes a rare dangerous adverse reaction right now? There are open questions with plenty of uncertainty. (I'm coming at this from the perspective of expert academic computer science, so I feel free to be equally honest about everyone's prospects, especially as my own field gets decimated. I'm a neuroscientist these days, explicitly because I hated what we were doing with AI, so I'm particularly interested in this field and other adjacent areas.)

I personally believe you need human-to-human contact for therapy.

It is very unlikely for psychiatry to be replaced because there is no emotional understanding and human connection in AI. AI can replace data analysis and probably chatbots but patients want empathy and trust. AI will make physicians more productive and allow them to shift their focus to providing human centered care. The best treatment will always need human connection.

High chance
most psychiatrists are already cold and are used to boxing and categorizing ppl which is literally what ai can do.

No matter how much Ai evolves, it can never be trained to accommodate everyone and their struggles

Well, AI therapists already exist - that is considered to be truth, as that they are already being used. But therapist is made to make you able to go back to the society, not leave it, and talkign with ai, which is raised to answer your questions to make you pleasured, and not necessarily speak truth, it will likely make you more and more anti-social, but yes - it might replace bad psychologists and therapists, who's best advice is to be happy that you made your bed this morning

Humans need to relate and have compassion from a real human and there is no chance no fake ai can complete that feeling.

Interpersonal skills, especially emotional intelligence are the key traits of both psychiatrist & the various counseling professionals. These are the hardest for AI to duplicate.

How could they it needs stuff that is near impossible to replicate

Chatbots for therapy already exist

Current GPT-4 already
"ranked higher than the majority of physicians in psychiatry, with a median percentile of 74.7% " on the 2022 Israeli board residency examinations than practicing physicians.

https://ai.nejm.org/doi/full/10.1056/AIdbp2300192

Artificial Intelligence in 20 years will categorically outperform psychiatrists in both diagnosis and prescription. Social frameworks for allowing AI to hold a diagnostic or prescribing role should have been implemented by then.

The amount of data and information that an AI can process cant be matched by us.
AI will remember each and every minute detail about the person once he tells it and by his past history data. AI will be able to suggest accurate strategies based on this. It might not happen suddenly but slowly slowly they will become more prominent.

Cause it needs a heart to heart connection and ethical skills like empathy and understanding which cannot be acquired by robots

because it will facilitate many tasks and allow you to have free time for other activities.

AI can provide non stop guidance and universal accessibility to a patient.

A profession which will be assisted by AI, although the main task will also be one whom will be preferred to be assisted by humans.

As there is always a small chance of something happening but still psychiatrist is a job which requires humans understanding other humans so most likely it would never happen

There's no way psychiatrists would be replaced/automated, their most important feat is to understand the emotions of humans, and that won't be able to be replaced by robots in the next two decades.

It is extremely unlikely that AIs will handle psychological issues with pinpoint accuracy. In order to treat a human, you have to be a human. Yes, codes and programs will eventually learn how to handle particular patients, but not all problems are the same.