When a company releases a new photo generator, it won’t be long before someone uses it to create a video of Will Smith eating spaghetti. It’s become a meme and a reference point: See if a new video generator can actually get Smith to down a bowl of noodles. Smith copied the model in an Instagram post in February. Will Smith and spaghetti are just one of the many strange “illegal” acts sweeping the intellectual world in 2024. Elsewhere, a British programmer has created a platform where AI can play games like Pictionary and Connect 4 against each other.
That’s not to say there won’t be more experiments in AI’s effectiveness. So why would the floodgates break? Or first, most industry-standard AI benchmarks don’t really show the average person. Companies often tout AI capabilities for answering math Olympiad questions or solving PhD-level problems. But most people (including you, actually) use chatbots for things like answering emails and basic research. Enter Chatbot Arena, a public test where many AI enthusiasts and developers are following the trend. Chatbot Arena allows anyone on the internet to test the effectiveness of AI on specific tasks, like building web applications or rendering graphics. But the raters appear unrepresentative (mostly from the AI and tech industries) and vote based on personal, hard-to-define preferences.
In a recent article on X, Wharton management professor Ethan Mollick pointed out another problem with most AI models in the industry: They don’t compare the efficiency of the process to the middle man.“It’s a real shame that there aren’t 30 different standards for different organizations in medicine, law, consulting, etc., because people are using this process to solve these problems,” Mollick wrote.Link 4, Minecraft, and weird AI criteria like Will Smith eating spaghetti are never realistic or even universally applicable. Just because Will Smith passed the test doesn’t mean he’s See More
0 Comments