Talk:Artificial intelligence: Difference between revisions
→Appeal posted re proposed deletion: new section |
→Scope?: add inaccuracy, bullet items to make clearer |
||
| Line 6: | Line 6: | ||
:I see AI more as a theme/background article. AI is so pervasive now, and affects people in so many ways, that I think it makes sense to have at least one article on it. | :I see AI more as a theme/background article. AI is so pervasive now, and affects people in so many ways, that I think it makes sense to have at least one article on it. | ||
:Things that I think such an article should cover include: | :Things that I think such an article should cover include: | ||
:Data centers - environmental impacts, community impacts, energy demand and subsidy by electricity and water rate payers, and how many of these agreements are made in secret, even in nominally democratic/open governmental systems. In the US data centers are often located in marginalized communities, where people are not as organized to protect their community . (This is not exclusively an AI thing might be worth a separate article about data centers in general, covering crypto mining operations, etc.) | :* Data centers - environmental impacts, community impacts, energy demand and subsidy by electricity and water rate payers, and how many of these agreements are made in secret, even in nominally democratic/open governmental systems. In the US data centers are often located in marginalized communities, where people are not as organized to protect their community . (This is not exclusively an AI thing might be worth a separate article about data centers in general, covering crypto mining operations, etc.) | ||
: | :* Inaccuracy and inappropriate use of LLM. "Hallucinations" People not understanding what an LLM is and assuming they are more capable than they are. LLM make a poor substitute for human written product reviews. (Inaccurate, praises whatever the user wants - even products that don't exist.) | ||
:Intellectual property - piracy in training data (using stolen data), use of output. | :* Control of information - Use of LLM in place of search is decimating independent information sources (taking away advertising revenue, taking away views). | ||
:Privacy and security - data poisoning, ease of subverting guardrails, gathering data for training, revealing prompts, law enforcement review of chatbot prompts and outputs, etc. | :* Intellectual property - piracy in training data (using stolen data), use of output. | ||
:Concerns about possible effects on users - AI psychosis, etc. | :* Privacy and security - data poisoning, ease of subverting guardrails, gathering data for training, revealing prompts, law enforcement review of chatbot prompts and outputs, etc. | ||
:Labor concerns - conditions of labelers/piece workers. | :* Concerns about possible effects on users - AI psychosis, etc. | ||
:Liability - LLM are often inaccurate, what happens when the AI harms people (libel, suicide, etc.) | :* Labor concerns - conditions of labelers/piece workers. | ||
:* Liability - LLM are often inaccurate, what happens when the AI harms people (libel, suicide, etc.) | |||
:I have sources for a bunch of this, will be adding them to the article talk page as time permits. [[User:Drakeula|Drakeula]] ([[User talk:Drakeula|talk]]) 18:27, 4 September 2025 (UTC) | :I have sources for a bunch of this, will be adding them to the article talk page as time permits. [[User:Drakeula|Drakeula]] ([[User talk:Drakeula|talk]]) 18:27, 4 September 2025 (UTC) | ||