what about meditationGPT?
People are going wild for ChatGPT out there. I probably shouldn’t add more noise to the conversation, but I’ve been asking it questions about meditation and just can’t help myself. Here’s a prime example:
Now, there’s certainly a lot of different ways to teach and practice meditation, but I think most of us would agree that ChatGPT dropped the ball on this one. The answer is impressive and even a bit useful, but it’s missing something essential.
A decent meditation teacher would have pointed out that silencing your inner voice is not the goal of meditation. A better response would have been to accept your mind as it is. Stop trying to resist thoughts. Let them come and go. Paradoxically, this might actually end up quieting your mind, but you can’t fixate on that. It’s not the point.
AI is getting crazy good at synthesizing information into coherent, organized answers, but it’s not going to question your assumptions. In this case, ChatGPT couldn’t do what meditation teacher and polymath Shinzen Young instructs his teachers-in-training to do before answering a question: “look for the question within the question”.
Speaking of Shinzen, when we hosted him to speak about technology at the Mindful Society Global Institute in 2021, he invited everyone to watch the documentary about AlphaGo. He then shared his vision for an AI-powered meditation coach. His prediction was certainly on point. It seems more plausible today than ever before. But let’s not get too excited: we’re not that close.
Sure, you could train ChatGPT to handle specific questions like this, training it with question-answer pairs from a wide range of contemplative paths. But let’s be clear - it is nowhere close to being able to do what my teachers have done for me.
Great teachers pick up on the subtle nuances of your experience, question your assumptions, and skillfully prompt you to eventually discover your own next steps. An AI serving up distilled information about meditation teachings and techniques would be amazingly useful, but let’s not confuse that with what human teachers can do.
—
Go chat with ChatGPT, and if its servers aren’t too overloaded, you’ll definitely get a sense of awe as it crafts eerily complete answers to both practical and playful questions. After deep interrogation, it will start to feel more like you’re talking to a highly advanced and creative search engine.
As an interface designer, I actually kinda love this. There’s something beautiful in ChatGPT’s honesty and transparency. If you ask a smart speaker “How are you?”, it will probably say something like “Good. You?” It will try to seem human, maybe even crack a joke. If you ask ChatGPT the same question, here’s what you get: “I am an AI and do not have the ability to feel emotions.”
ChatGPT is not trying to be human and failing, it’s trying to be useful and succeeding. If you try to Google a specific issue, often you’re stuck going in and out of various websites trying to find themes. It takes a while to sift through the noise and it’s difficult to know who to trust. ChatGPT does this for you and delivers a cohesive and organized answer in natural language that feels trustworthy, even if it omits its sources and censors offensive content with the help of underpaid workers in Kenya.
We see similar patterns with MidJourney and DALL-E: the vanguard of AI tools that create art. Well, I don’t know if we can call it ‘art’. It’s more like a hodgepodge of our visual ideas layered and mixed in surprisingly coherent ways. It fluently synthesizes a massive amount of existing images into slightly awkward yet shockingly attractive images tailored to your specific prompts.
Although the output is superficial, these AI image generators are useful. They combine existing styles in ways a human wouldn’t, and so they definitely do produce new ideas. But just as ChatGPT blindly accepted my meditation question, visual AIs often fail to challenge assumptions. They consistently reproduce human biases in some pretty problematic ways.
This new wave of AI has produced evocative new user interfaces for research, creative writing, art, design, and more. But to the worried artists, writers, and meditation teachers out there: you can relax. We still need you just as much as we still needed musicians after synthesizers exploded in the 80s.
Maybe we even need you more than we used to. As AI automates superficial tasks, I hope it helps dissolve the production mentality around art. I’m not too concerned about outsourcing clickbait listicles and bland hotel art to AI, as long as we also see an uptick in how much we value unique human abilities that depend on conscious experience.
An AI meditation teacher will save you a ton of time gathering information, but it won’t truly understand human experience the way great mindfulness teachers do. And as it turns out, that’s what great writers and artists do, too. Our tools are getting amazing, but they’re still just tools.