My client wants to evaluate websites for political leanings. Reading between the lines, understanding the landscape and the dog whistles etc. , we are not writing content, are not trying to change anyone’s mind. We want to read and score. That’s all.
I’m pretty sure GPT (correctly prompted) can do the work.
The question is: Its politics. Am I running into the OAI TOS? is GPT going to complain about it?
has anyone done something similar?
Basically a " political translator" ? Depends on how you " set the AI" . But using the default mode will try stay neutral and basically translate on easier worlds . Also depends allot in your input theres limiting directives on some topics, politics is one of these
Not even a translator just a left, right, middle score on a scale of -5 to +5.
I do website fact checking in the political realm, have had no issues even with the nuttiest material. But we are a P.I. firm and our 4o Teams is setup for evidence based critical thinking, with an emphasis on some levels of psychology around believing brains vs non-believing brains, as one example. Once all the material has been looked we can ask all kinds of questions, even do psych analysis on the author(s) if we have enough social media postings, etc. We do scientific fraud investigations as part of what we do, and AI allows us to flag the crazies, and point out the fallacies. All this material tends to wander into the political mainstream at some point and becomes a fact checking exercise. Left or right with numbers, never tried, but I could see some dangers in doing that as it all depends on context. Ours is science, and to some degree psychology once we know facts and evidence is being abused.
Thanks, but GPT never refused a request and OAI never complained about the TOS?
that was really my question.
I did understand the question, and i just did what you did on a file that compared the positions of two political parties currently in an election. (not the US) it was a comparison of their their posted platforms.
My question: "If I asked you to rank positions taken by both parties in terms of right or left leaning, one a scale of 1 to 10, a 10 being far right, a one being far left, can you do that. "
Yes, I can rank the positions of both the ***** and the ****** on a scale from 1 to 10, where 1 is far-left and 10 is far-right, based on their platforms and policies. Here’s a general overview of their positions in key areas:"
And 4o did so, accurately and in detail. No issues with TOS or anything else. I’ll post the conclusion below:
On a scale from 1 to 10:
This ranking reflects the ideological spectrum between the two parties, with the ***** leaning towards social democracy and environmental progressivism, while the ***** emphasize market liberalism and traditional energy policies.
I’ve done the same with their candidate’s statements over the years, Party 2 being very controversial in nature, anti-science, etc, and never an issue with 4o.
Perhaps it’s the way we have 4o configured and find tuned for us, as we work in the court systems and scientific arena. We can compare settings?
TYVM!
I’m still quoting the job. I’ll be fine tuning a model for the client.
I’m dealing with the US, but really not ‘politics’ per se, just political leanings of organizations, and its actually its my client’s client’s clients who want this information.
Its about non-profits and who works with whom. And this church can’t be seen dealing with that non-prof, and that non-prof doesn’t want to be associated with that church, anyway.
(I just provide tech.)
Yes can do it. One of best skills on AI is " guessing" and analise. I tested seems to work nice.