A cautionary tale for those planning festive swims: the perils of relying on AI for tide advice.
As thousands of people prepare to take the plunge in UK waters this festive season, a timely warning from the Maritime and Coastguard Agency serves as a reminder that AI tools can make mistakes.
Two individuals learned this the hard way after becoming stranded on Sully Island near Barry, Wales. ChatGPT, an AI language model, provided incorrect tide times, leading to a rescue operation by the coastguard.
This incident echoes the words of Sundar Pichai, CEO of Google's parent company Alphabet, who recently cautioned against blindly trusting AI models, highlighting their susceptibility to errors.
One of the rescued individuals, Alan Hughes, shared his experience on social media, admitting to using ChatGPT for research on low tide times. The AI program's response was off by two hours, leaving the causeway underwater when they attempted to return.
"Some false information there, but maybe a lesson learned for me," Hughes reflected.
HM Coastguard emphasized the importance of using reliable resources, such as the UK Hydrographic Office's Easy Tide or the Met Office's service, for accurate tide information.
When the BBC tested ChatGPT with the same query, it generated the same incorrect response, highlighting the need for caution when relying on AI for critical information.
Open AI, the owner of ChatGPT, acknowledged that improving factual accuracy is a significant focus for the industry, but hallucinations (providing incorrect or fabricated information) still occur.
Prof. Steven Schockaert from Cardiff University's School of Computer Science suggested that the issue may lie with how information is presented online, particularly in tables, which can be challenging for language models trained on left-to-right text.
"I think it's probably not a good idea to rely on AI for information that you can't verify and that's somehow important, like in this case, the tides," Schockaert advised.
Sam Skyrme-Blackhall, chair of the Tenby Sea Swim Association, echoed these sentiments, emphasizing the importance of listening to professionals and local experts when it comes to understanding the challenges of coastal waters.
"These are tidal beaches. I check the tide times weekly, but I also check the conditions. Listen to the people who have the knowledge," Skyrme-Blackhall urged.
AI's impact on outdoor adventures is a double-edged sword. While it can provide valuable assistance, it can also lead to dangerous situations if not used wisely.
Aberdyfi Search and Rescue Team rescued two walkers who had used AI to plan their excursion up Cader Idris near Dolgellau. Team leader Phil Britton and deputy leader Gareth Davies, with over 50 years of experience between them, noted that the walkers were well-prepared but lacked the necessary judgment due to their inexperience.
"In mountain rescue, we find that there's all the information there, but it's always interpreting it to the real-life situation if you haven't got the experience," Britton explained.
Despite the potential pitfalls, AI can also be a lifesaver. In the case of the walkers on Cader Idris, AI provided the necessary guidance when they realized they were beyond their capabilities.
"An error in judgment" is how Gareth Davies described the situation, emphasizing the importance of human experience and local knowledge.
"Technology's brilliant, but you've got to know its limitations," Davies concluded.
So, as you plan your festive swims or outdoor adventures, remember to exercise caution and seek reliable sources of information. While AI can be a powerful tool, it's crucial to understand its limitations and not rely solely on its advice, especially when safety is at stake.
