I don’t think we should trust Chinese AI for the same reason we shouldn’t trust TikTok. There are plenty of other AI options that aren’t in bed with a genocidal slave state.
Robin -- thanks for this great response. I was just looking at your posts and stack. Will keep following and reading you, especially as you dig into the visual tools!
I also messed around with Deepseek last night, then ChatGPT this morning. I was interested in seeing the thresholds associated with writing two-liner jokes and more. Both models churned out predictable results (Cheesy, mediocre dad jokes, there's likely a pun here that I'm not digging for). However, ChatGPT this morning was willing to push the envelope for more edgier responses, to include more verbose anecdotal jokes. DeepSeek continued to refer to its preset guidelines to avoid producing a potentially harmful response and after I began cutting and pasting responses, it essentially timed my responses out.
I think DeepSeek has been a forkable project via GitLab for awhile now, so I am curious to know if any content creators or screenwriters will be cloning and modifying the source code to assist with draft development. I see a utility for a tool like this to assist treatments and specs, albeit the end result will most often always be some form of a template.
It was cool to see DeepSeek push out .py scripts of games and other simple programs and I imagine this could be done for other languages as well. It's an interesting time to be alive.
Dear Fred, thank you for sharing your perspective and experience with us so openly. It truly helps to gain some insight.
Also, thank you for pointing out the Jevons Paradox! This is the second time I’ve come across this effect, which I hadn’t heard of before, but it clearly makes sense and has occurred often in history. That provides some useful perspective.
I don’t think we should trust Chinese AI for the same reason we shouldn’t trust TikTok. There are plenty of other AI options that aren’t in bed with a genocidal slave state.
Robin -- thanks for this great response. I was just looking at your posts and stack. Will keep following and reading you, especially as you dig into the visual tools!
I barely post anything as I've been in the Davinci Resolve cave for several weeks now ;) Appreciate the follow,
Cheers Fred,
I also messed around with Deepseek last night, then ChatGPT this morning. I was interested in seeing the thresholds associated with writing two-liner jokes and more. Both models churned out predictable results (Cheesy, mediocre dad jokes, there's likely a pun here that I'm not digging for). However, ChatGPT this morning was willing to push the envelope for more edgier responses, to include more verbose anecdotal jokes. DeepSeek continued to refer to its preset guidelines to avoid producing a potentially harmful response and after I began cutting and pasting responses, it essentially timed my responses out.
I think DeepSeek has been a forkable project via GitLab for awhile now, so I am curious to know if any content creators or screenwriters will be cloning and modifying the source code to assist with draft development. I see a utility for a tool like this to assist treatments and specs, albeit the end result will most often always be some form of a template.
It was cool to see DeepSeek push out .py scripts of games and other simple programs and I imagine this could be done for other languages as well. It's an interesting time to be alive.
Dear Fred, thank you for sharing your perspective and experience with us so openly. It truly helps to gain some insight.
Also, thank you for pointing out the Jevons Paradox! This is the second time I’ve come across this effect, which I hadn’t heard of before, but it clearly makes sense and has occurred often in history. That provides some useful perspective.