Skeptics Guide #885

The Skeptic’s Guide to the Universe Skepticast #884 June 21st 2022 Segment #1. Special Segment Secret of Skinwalker Ranch Segment #2. News Items News Item #1 – Jupiter Ate Baby Planet https://www.livescience.com/jupiter-ate-baby-planets-while-growing News Item #2 – The Risk of Sitting https://www.sciencealert.com/around-the-world-people-are-dying-too-soon-because-they-sit-around-too-much News Item #3 – NASA Joins Study of UAPs https://theness.com/neurologicablog/index.php/nasa-joins-study-of-uaps/ News Item #4 – Galapagos Giant Tortoise Not Extinct https://www.nature.com/articles/s42003-022-03483-w Segment #3. What’s the Word Nostrum Segment #4. Your Questions and E-mails Question #1: Testing AI Hi SGU, I love the podcast. In Episode 884 you had a really interesting discussion about LaMDA, the Google chatbot, and whether it was sentient. I agree with your conclusions but found two of your lines of reasoning problematic. Firstly, this idea that the emotions being displayed were "too human". If the chatbot is being trained on human emotions, it is unsurprising to me that a sentient AI would use those human words and terms to express themselves because that is the only language it has available. They are trying to create something to sound human, so we shouldn't be surprised that is how that chatbot talks, sentient or not. But more broadly, I struggled with the logic of "this is not how I imaging it would happen or look like, therefore it is not likely to be true". Our preconceptions could easily be wrong. I certainly believe this is a logical fallacy being deployed. And whilst informal logical fallacies are informal and therefore sometimes fine to use, in this case, the first potential AI, I think it is relevant. We have never seen a real true sentient AI. Therefore your own assumptions of what it should look like carries even less weight. All this to say, the engineer clearly fell platonically in love with a piece of software, like the chao pets in Sonic the Hedgehog from the Dreamcast era for many computer game players. He failed to actually properly interrogate the software. LaMDA is not sentient but is impressive. If you believe I have misunderstood your positions or believe my own logic/reasoning is faulty do let me know please. Best regards, David (Leeds, United Kingdom) Segment #5. Science or Fiction Each week our host will come up with three science news items or facts, two genuine, one fictitious. He will challenge our panel of skeptics to sniff out the fake – and you can play along. Theme: Geology #1) Geologists estimate that the next major earthquake along the San Andreas fault would cause vastly more damage from the resulting tsunami than the quake itself. #2) The Pacific ocean is so large that it contains its own antipode. #3) Sapphires and Rubies are actually the same mineral, corundum. Segment #6. Skeptical Quote of the Week "If we're never wrong then we're never surprised. If we grow too protective of our existing beliefs, then we stagnate and stop learning. If we've reached any degree of competence within our field, it's because we got things wrong along the way. So why stop now? It's interesting, I think, for each of us to consider the following: What am I currently wrong about? It's an impossible question to answer, but a curio though to contemplate nonetheless." - Hector Chadwick


View on YouTube.