There is nothing particularly new here with AI, except its name.
Back in the mid 70s I had the opportunity to play with an early language model developed at MIT, Eliza. Eliza was designed to understand how human language worked and was understood. Using a keyboard the human started a conversation and Eliza responded using reflective techniques and common social conversations. Due to the technology available in the mid 70s, Eliza was limited in its ability, however experiment with humans often had the humans believing it was all a ruse and there was a human in another room answering. My experience was similar, I became frustrated at Eliza's inability to really understand what I was trying to say, even though I knew it was a machine.
The large language models are based on neural networking theory and probability statements. Back in the 80s I considered writing a dissertation on neural networking models of reading. The theories were pretty interesting and based on probabilities, the gist being that letters have higher probability of following certain letters and not others. For example, there is very low probability of x following c in a word in English (I think the probability is 0) while there is a high probability of L following C. The large language models do basically the same thing, it analyzes lots of written material, calculating the probability any one word following any other word.
The kind of model
@Davidasailor26 mentions has its roots in early 20th century agriculture research laid the ground work for multivariate statistical analysis. Multivariate analysis looks to see how different factors are related. In early agricultural research, scientists would look at just a few factors, such as soil type, fertilizing protocols, seed type, and so forth. David's suggestion of customized polars for his boat is nothing particular fancy, however, it can be extremely data intense. Highly competitive sailors have been doing it for years, however the data measurement s have been crude and the number of data points a human brain can handle is much smaller than that which can be handled by today's technology.
In the end, all AI is basically formulaic, original ideas, novel interpretations, and nuances are just not present in today's AI world. AI is just a set of rules for data interpretation and regurgitation. In recent months there have been posts made here on SBO that seem to have all the hallmarks of a ChatGPT answer. The formula is pretty simple and easily recognizable, it starts with an introductory statement or two about the problem, something like "Many sailors have had similar issues with....Some of them have found these things to be helpful" and then there is a list of possibilities are vague and non specific. And finally a summary statement that says basically nothing.
Just for grins I used John's opening statement for this thread for a ChatGPT prompt. Here's the unedited response.
AI is indeed poised to make a significant impact across a wide range of fields, including sailing. As technology continues to advance, its integration into sailing could reshape the way we approach and experience this activity in several ways:
1. **Navigation and Route Planning**: AI can greatly enhance navigation by analyzing weather patterns, sea conditions, and traffic to optimize routes. It could help sailors avoid hazards, choose the best routes, and even adjust course in real-time based on changing conditions. This could be particularly valuable for long-distance cruisers and competitive racers who need to make precise decisions.
2. **Training and Skill Development**: For new sailors, AI-powered apps could offer personalized training programs, tutorials, and simulations to accelerate learning. These tools could provide instant feedback and adaptive learning experiences, making it easier to pick up sailing skills and improve techniques.
3. **Autonomous Sailing**: AI could potentially take over some aspects of piloting, especially for routine or repetitive tasks. Autonomous or semi-autonomous sailing systems could handle navigation, sail adjustments, and even docking, allowing sailors to focus on other aspects of their journey or simply enjoy the experience.
4. **Safety and Monitoring**: AI could enhance safety by monitoring the boat's systems, detecting potential issues, and providing alerts. It could also analyze data from sensors to predict and prevent mechanical failures or other problems before they become serious.
5. **Performance Enhancement**: In racing, AI could be used to analyze performance data, provide strategic insights, and even simulate different race scenarios to help teams make informed decisions. This could give competitive sailors a technological edge, refining tactics and optimizing boat performance.
6. **Personalization**: AI could tailor the sailing experience to individual preferences, adjusting settings for comfort and convenience, recommending destinations based on interests, or providing insights into local conditions.
Overall, the integration of AI into sailing holds the promise of making the experience safer, more efficient, and more enjoyable. However, it also raises questions about reliance on technology and the potential for privacy concerns. Balancing the benefits of AI with careful consideration of its implications will be key to making the most of these advancements in sailing.
Kind of looked like it paraphrased this thread. Maybe I should try this again after it has a chance to review this post. Anyway, you can easily see how formulaic the response structure is.
What do you think we should do to people on the internet that simply use AI to answer questions in a forum?