I have been warning against AI for one specific area. Today, reading yesterday's Daily Mail, I live in Portugal so they arrive a day late, I came across the same warning in the paper's Good Health section. Now, whatever the limitations of the newspaper, in this section it can be outstanding. The writer, Daniel Oberhaus, opens up with his younger sister who killed herself after extensive bullying at school and psychiatric help, which worsened her feelings. Her attempts to explain herself to psychiatrists told her only that they lacked genuine knowledge. The writer, a science journalist, echoes my position: It is a medical area that has failed to genuinely find biological origins for the 'diseases' it claims to be expert in.
Now, psychiatrists, wholly against the evidence and after 70 years of prescribing dangerous, toxic drugs as treatments, thereby expanding mental illness, claims to be a science. No adequate treatments! Just their claims. But the writer, while agreeing with my estimate, is aghast at the introduction of AI in the process of diagnostic activity through technology and the likelihood of millions being judged mentally ill as a result of the rhetorical nature of the 'science'- Many lives will be ruined and brains ruined by the dangerous drugs employed. Psychiatry is a science because they, the psychiatrists, not the evidence, tells us it is.
Since I became informed on AI, I have been immensely worried on this matter as AI does not really think but functions on the evidence out there.
Can anyone see an additional problem given my final statement. AI is a danger if the statements it is given are false......
Your concern that AI will replace psychiatrists and use only a limited framework in diagnosing people is valid. Psychiatrists are already doing that, in part, because they prescribe drugs, limiting their diagnostic options. (Psychologists provide counseling and are more willing to explore creative options.) While AI might someday evolve to provide highly individualized treatment, its current limitations would only exaggerate the tendency of psychiatrists to classify patients based on the drugs that are available.
Keith, certainly my concern but unfortunately its not true anyway. Psychiatrists are certainly driven by the pharmaceutical companies, which actually predict new illnesses.
But in addition, diseases change or can change, simply evolve, and any information on diseases employed now by AI could become irrelevant in the future.
I think that AI should not be used as an only agent in sensitive and critical scenarios like medical treatment if the models can't themselves describe what it is doing inside or how it is working internally. Currently, we are living in an era where AI is a hype and we are applying AI in almost every scenario we know, without knowing how these systems would explain themselves. I think this is the time when we need to focus more on Explainable AI (XAI). One of the dangers of the AI models is implementing these models in situations without the ability of XAI, or more specifically implementing deep learning systems in real-life scenarios without the DL model having the power to explain themselves.
I sometimes feel worried that people are creating AI and ML based products that they claim to behave like humans. Many people claim that AGI or Human Like intelligence will be achieved soon. Regarding these things I just have one question, we still haven't figured out how human consciousness evolves or works, then how can we create AGI or Human Like intelligence artificially where consciousness plays a very crucial factor?
I think these are some of the dangers that are being overlooked currently.
You are right if given a false evidence or wrong statement, AI doesn't have a thorough intuition on what to prescribed for a psychiatric patient except what is program into it which maybe contextual biases!
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Materials Science
- Materials
- Composites
- Nanocomposites