This reminds me of many years ago when I worked in radio doing news. I always struggled reading foreign names or even just difficult American names.
The news director, who was somewhat of a local legend, gave me a great piece of advice that I still remember. He said "say the name authoritatively and say it the same way every time, and people will believe it's correct."
That sounds a lot like the response from Claude. Only my advice came from a person back in 1982.
This is written like Claude has feelings that influence how it works. I think the truth is is that the coders prioritized these shortcuts. At the same time there is no way around error because Claude is pulling from Human information which contains error. It chooses answers based on what humans would most likely say. So its coded on popularity rather than accuracy. This could get into a whole conversation around philosophy and epistemology haha. But basically, you cannot program an AI to be more accurate than the info that it is pulling from. AI cannot transcend humanity's limitations. It can only assist us with left brain thinking so that we can focus my on right brain production/thinking.
People who use AI are lazy. Claude admitting that it's lazy and just giving a quick answer that sounds good, and written in a very confident voice, is the most on-brand thing ever.
People who use it with no checks and balances in place are making a mistake, but if it frees up your time to do higher value work, that to me is working effectively and efficiently versus laziness.
Very interesting and informative article. For whatever reason when I read "Make me show my work" I immediately thought of this cartoon. https://share.google/N2jtFwqxhYMvHVDos
This reminds me of many years ago when I worked in radio doing news. I always struggled reading foreign names or even just difficult American names.
The news director, who was somewhat of a local legend, gave me a great piece of advice that I still remember. He said "say the name authoritatively and say it the same way every time, and people will believe it's correct."
That sounds a lot like the response from Claude. Only my advice came from a person back in 1982.
President Claude - Making AI Great Again
That’s funny. I think it works too if you decide to run for President.
Sadly, it does.
This is written like Claude has feelings that influence how it works. I think the truth is is that the coders prioritized these shortcuts. At the same time there is no way around error because Claude is pulling from Human information which contains error. It chooses answers based on what humans would most likely say. So its coded on popularity rather than accuracy. This could get into a whole conversation around philosophy and epistemology haha. But basically, you cannot program an AI to be more accurate than the info that it is pulling from. AI cannot transcend humanity's limitations. It can only assist us with left brain thinking so that we can focus my on right brain production/thinking.
People who use AI are lazy. Claude admitting that it's lazy and just giving a quick answer that sounds good, and written in a very confident voice, is the most on-brand thing ever.
Have to disagree that people who use AI are lazy.
People who use it with no checks and balances in place are making a mistake, but if it frees up your time to do higher value work, that to me is working effectively and efficiently versus laziness.
Very interesting and informative article. For whatever reason when I read "Make me show my work" I immediately thought of this cartoon. https://share.google/N2jtFwqxhYMvHVDos