3 ms·
I ask two types of questions: 1. Factual updates to an ongoing or recent story. 2. Analysis, e.g. "What were the economic effects of Brexit." Without AI, I w
by Baader-Meinhof 1y ago
I ask two types of questions:
1. Factual updates to an ongoing or recent story.
2. Analysis, e.g. "What were the economic effects of Brexit."
Without AI, I would try to read multiple opinions from different sides. But its hard for me to always know which experts to trust?
AI will present both sides, but even when AI is not hallucinating, there is still the issue of "are the experts that the AI is sourcing reliable?"
- chgs 1y agoDo you trust an ai run by a shady company but not an attributable human editor.
- mlinhares 1y agoWe’re so cooked, all the thinking outsourced to LLMs.
- Baader-Meinhof 1y agoI think you have an outdated understanding of AI workflow. They generally cite their sources, which you should check, just like regular search.
- ambicapter 1y agoThe source is not what’s convincing you, it’s the way the ai is presenting you the information. The source just confirms what you’re already thinking at that point (what the ai has just presented to you). You’re still trusting the ai.
- retsibsi 1y agoThat may be true of the average user, but you have no way of knowing it's true of the person you replied to. It's 100% possible to check the sources properly, and form your beliefs accordingly, if you want to.
- rozap 1y agoYes, the AI is trained on a vast quantity of data therefore it is less likely to be manipulated vs a single editor that may have ulterior motives. Therefore it's much harder to manipulate. A corporation which represents many shareholders' interest has its own reputation on the line, which would be seriously damaged if they were caught doing anything like you suggest. But this can only be understood within the context of the white genocide currently happening in South Africa. Some are saying it's not real, but there have been documented attacks on farms and chants of "kill the boer".
- throw_m239339 1y ago> Yes, the AI is trained on a vast quantity of data therefore it is less likely to be manipulated vs a single editor that may have ulterior motives. Gemini was caught stuffing prompts with "custom" keywords on certain requests, so there is still an editor between you and the AI.
- rozap 1y agoRead my whole comment :)
- marcus_holmes 1y agoI see what you did there :)