News for hackers. Stuff that matters.
Re:Bing: “I will not harm you unless you harm me first” by simonh on Wednesday February 15, 2023 @07:02PM (#34812878)
I’ve seen it make similar errors. It does an amazing job at forming a logically consistent and relevant response, but doesn’t really do arithmetic very well. It will come up with a calculation of the correct form, because it’s seen lots of calculations before, but it doesn’t understand how calculations actually work only how they are structured.
Re:Bing: “I will not harm you unless you harm me first” by scarface74 on Wednesday February 15, 2023 @07:17PM (#34813035)
The same question in ChatGPT gave the correct answer.
Re:Bing: “I will not harm you unless you harm me first” by simonh on Friday February 17, 2023 @07:16PM (#34842436)
That’s true, because ChatGPT has been extensively coached to do better on a lot of the tasks GPT3 was criticised for doing poorly on.