In today’s column, I will be diving further into the realm of using generative AI for mental health. This is a vital topic and one that I’ve been covering on an ongoing basis.
Well, please do not be so quick to disregard ELIZA. As you will soon see, there is a lot of life left in caring about how it worked and what it did. I will take you into some interesting details and reveal the underlying magic of sorts. Next, I will provide you with another legendary head-to-head comparison by taking a mindful look at a dialogue that took place when ELIZA and PARRY were pitted against each other. Doing so will further reveal how they work and the kinds of AI techniques being employed.
AI makers typically undertake a concerted effort to try and reduce the potential for toxic outputs. For example, they often use a technique known as RLHF , before releasing their generative AI to the public. This process involves hiring humans to examine various outputs and indicate to the AI whether there were things wrong with those outputs such as perhaps showcasing biases, foul words, and the like.
A strong dose of healthy skepticism and a persistent mindset of disbelief will be your best asset when using generative AI. Who has the legal liability regarding those people harmed using generative AI for mental health purposes? The program was developed by a legendary computer scientist known as Joseph Weizenbaum. He was born in 1923, wrote ELIZA when he was in his 40s, and sadly passed away in 2008. His ELIZA program especially reached notable attention in 1966 when he published a showcase article about how the code worked.The ELIZA program made use of early-on AI techniques. Those AI techniques can be roughly lumped into the AI category of Natural Language Processing .
Scripts had to be written by people. Those scripts were fed into ELIZA. Whatever you saw ELIZA doing was driven by the human-devised script. It was the human that brought to the table a script that made ELIZA appear to be exhibiting intelligence. A comparison at times has been made to how Oppenheimer later in life had reservations about his work on the atomic bomb. Some even argue that if AI is truly an existential risk, see my coverage on the topic atAnyway, for purposes here, I want to get you up to speed on how ELIZA works. This will be instrumental for my discussion showcasing examples of ELIZA.