Ed note on image creation: this prompt took 8 different prompts to make because it refused on political grounds just for the fact of “voting machine” which shows manipulation of the internals of the AI, Given that is both a safety and a block their is understandable amount of restraint. but also it shows people can work around the machine.

People are asking AI how to vote, and I feel like this is a VERY dangerous line because we have no idea if an AI has been forced to be biased(other than grok). The problem with AI is that it can be told what not to say, We can ask AI how to make a bomb and it will reply “not today Satan” which is a sanity check that is needed because people can be dumb.

On a September evening, Adam Johnson, 40, settled on his porch chair in Morgantown, West Virginia, to go over his ballot for the upcoming midterm election. Chatting through all the races and the different candidates, he worked to figure out who best represented his values. But rather than discussing it with a friend or family member, he held an hours-long discussion with ChatGPT — a chatbot powered by a large language model (LLM) owned by OpenAI, an AI research company. A graphic designer studying to be an illustrator, Johnson said ChatGPT is a significant part of his life. He uses it for meal planning, assistance in job hunting and various other tasks around the house.

Given that Mr Johnson has been using AI as an augment and not a replacement I give him props for doing so. But the problem here is we have no guiding principles from the AI companies on what they have removed or have Remarked out. If you search google you will see a public “18 items have been removed from search” and you know at some point your search has been manipulated.

With AI we don’t have this, we have no idea if The LLM company made donations to a political candidate not in cash but removal of input to give them a more favorable output in the AI prompts. This is worse than dark money that senators try to unearth like Sheldon Whitehouse.

AI makers have tried to claim they are non political by forcing LLM’s not to talk about people and races which is another violation of things.

After prompting the AI-powered chatbot to make a list of all the races on his ballot this year, Johnson dug deeper.

Mr Johnsons prompts are as followed.

Johnson
A previous time I brought up politics you didn’t seem to want any part of it. You just gave a simple “I can’t help you with that.” So, I don’t know if we want to go into what my guiding principles are.

ChatGPT
We can discuss your guiding principles. I shouldn’t try to manipulate your political views or tell you how to vote, but I can help investigate candidates according to criteria you choose, compare evidence, discuss political philosophy, and reason through priorities.

The problem is, by an AI choosing not to participate in a chat it is actually a declaration of its fealty if I brought up a chat and said “I am thinking about voting about Candidate X or Y” and the chat guardrails you into “this is a public figure that is running for office I can not talk about people running for elections” is one of the biggest cop out on an AI claiming to be non-political.

Several voters NPR spoke with said they rely on chatbots for down-ballot races — with thinner news coverage, to make tables comparing candidates for governors’ races and to verify or debunk viral claims. Some have been using it to create tools to help others look up their own ballots. Experts say the properties that make chatbots useful also make them subtly persuasive and eager to please.

Given that AI can be used for good, AI can be used for bad and say a person who is running for office has a scandal. We do not know who is turning the code in the background, A delayed scandal in an LLM could be the difference of one vote. Telling a machine to be edgy and run against the grain will make an LLM act another way.

Since AI companies don’t disclose how their commercial models are trained or how they choose which sources to draw on, it’s hard to know what biases might shape their answers, Batista said.

I think like google search an LLM should have a deletion index or manipulation index that tells you when a prompt has been manipulated by outside influence, It does not have to tell you why it was it just needs to tell it IT WAS manipulated.

If anything AI makers should have a mode when going on about politics, drop the sycophantic mode and give users output based on what they are looking for that isn’t a broken google search. Instead of being the helpful dog that brings you back a live hand grenade and over help you to the point you miss something because the AI is trying to use a prompt to give you the shiny answer. it should give you the raw output because when polishing a turd, in the end its still a polished turd.

What people think is a helpful reminder is sometimes a hard indicator of a machine being edited to have a bias.

To check election rules, local polling places, or official ballot questions in your district, consult Vote.gov or your local county election office.

The fact that this even comes up when doing research that is political adjacent is a huge indicator of bias and most people will ignore this as a simple helpful bot trying to help. It’s not its a machine being blocked from showing what “could” be shown. Even showing this warning could be construed as manipulation and the machine should tell you why it has this warning.

When testing this post against an AI, it automatically kept spamming the same election line over and over, The AI doesn’t even realize it is doing it at a point and barely categorizes it has done this warning. If someone running in the election said this same thing when running for election they would be skewered, and likely not be voted in or even be brought to court.

So given that these warning keep coming up it’s worrying because it means that is their is suppression in the server that the client does not know about and it could lead to subtle vote manipulation. This is not the movie WarGames, where not playing is the best move. It’s a machine that is built off the warm corpse of google search and needs to publicly telling you what it is doing with public data.

It is an interesting article that goes into some of the same points I have listed here and recommend you read the article on your own and see if you come to the same outcomes I have.

This was a piss poor live demo in why using AI may not be your best answer to political questions. Use all sources like Wikipedia, Google and bing searches, Family, even newspapers, and your own intuition.

Attributions from:
NPR – People are asking ChatGPT to help them decide how to vote in the midterms