Economic concerns and growing disenchantment with both parties is draining support for Trump among Gen Z young men, a key bloc of support during the 2024 election
Male Gen Z voters are breaking with Donald Trump and the Republican party at large, recent polls show, less than a year after this same cohort defied convention and made a surprise shift right, helping Trump win the 2024 election.
Taken with wider polling suggesting Democrats will lead in the midterms, the findings on young men spell serious trouble for the Republican Party in 2026.
Younger Gen Z men, those born between 2002 and 2007, may be even more anti-Trump, according to October research from YouGov and the Young Men’s Research Project, a potential sign that their time living through the social upheavals of the Covid pandemic and not being political aware during the first Trump administration may be shaping their experience.
have the AI cite it sources for the claims it makes, and read through those sources. ask the AI what information it used in the source to come up with its statement.
using chatGPT in this way is like how I used Wikipedia growing up: just pull up an article on something, check out what citations an article had, and walk it backwards from there
When searching something online, I already many times came across AI written articles being quite high. So now it’s possible LLM will just cite another LLM which got its content from an LLM written response on Reddit.
Does the paid chat gpt actually cite?
The free one (which most people are using), when asked how much an aircraft carrier costs just links to the main page of CNBC and the department of defense.
I further prompted and it gave me 404s
It does a mix of hallucination and self-correction once you tell it to search the web to actually find out current information. And even then, you don’t have a good chance of having truly accurate information. It really does take work.
not sure if only the paid gpt actually cites, but anyone thats trying to show someone else how inefficient LLMs are with data should be able to come to the conclusion the output is untrustworthy if you’re getting 404 errors after asking to cite, in my opinion
The problem is it “cites” things by inserting obscured hyperlinks that people see and never click but assume there is a “source”
I guess we found the answer to how the original person i was replying to can validate anything thats generated by an LLM: looking up sources that an LLM claims to use