{"id":286906,"date":"2023-04-10T06:51:11","date_gmt":"2023-04-10T10:51:11","guid":{"rendered":"https:\/\/knowtechie.com\/?p=286906"},"modified":"2023-07-04T02:27:40","modified_gmt":"2023-07-04T06:27:40","slug":"chatgpt-and-other-language-ais-are-just-as-irrational-as-we-are","status":"publish","type":"post","link":"https:\/\/knowtechie.com\/chatgpt-and-other-language-ais-are-just-as-irrational-as-we-are\/","title":{"rendered":"ChatGPT and other language AI’s are just as irrational as we are"},"content":{"rendered":"

The past few years have seen an explosion of progress in large language model artificial intelligence systems that can do things like write poetry<\/a>, conduct humanlike conversations<\/a> and pass medical school exams<\/a>. <\/p>\n\n\n\n

This progress has yielded models like ChatGPT that could have major social and economic ramifications ranging from job displacements<\/a> and increased misinformation<\/a> to massive productivity boosts<\/a>.<\/p>\n\n\n\n

Despite their impressive abilities, large language models don\u2019t actually think. They tend to make elementary mistakes and even make things up<\/a>. <\/p>\n\n\n\n

However, because they generate fluent language, people tend to respond to them<\/a> as though they do think<\/a>. <\/p>\n\n\n\n

\"ChatGPT
Image: Pexels<\/figcaption><\/figure>\n\n\n\n

This has led researchers to study the models\u2019 \u201ccognitive\u201d abilities and biases, work that has grown in importance now that large language models are widely accessible.<\/p>\n\n\n\n

This line of research dates back to early large language models such as Google\u2019s BERT, which is integrated into its search engine and so has been coined BERTology<\/a>. <\/p>\n\n\n\n

This research has already revealed a lot about what such models can do and where they go wrong.<\/p>\n\n\n\n

For instance, cleverly designed experiments have shown that many language models have trouble dealing with negation<\/a> \u2013 for example, a question phrased as \u201cwhat is not\u201d \u2013 and doing simple calculations<\/a>. <\/p>\n\n\n\n

They can be overly confident in their answers, even when wrong. Like other modern machine learning algorithms, they have trouble explaining themselves when asked why they answered a certain way<\/p>\n\n\n\n

\n