“People tell chatbots about their medical fears, their relationship problems, their beliefs about God and the afterlife.
-
@Em0nM4stodon why all of those announcements feels like marketing?
@Lioh @Em0nM4stodon yeah this article is still very boosterish in a lot of ways.
This person quit the company, but they aren't done detoxing from the hype yet.
-
“People tell chatbots about their medical fears, their relationship problems, their beliefs about God and the afterlife.
Advertising built on that archive creates a potential for manipulating users in ways we don’t have the tools to understand, let alone prevent.”
https://globalnews.ca/news/11664538/anthropic-ai-safety-researcher-mrinank-sharma-quits-concerns/
@Em0nM4stodon We HAVE a brain. It understands very well. It doesn't have to know all the specific ways it manipulates, just extrapolate from algorithmic (?) manipulation.
-
“People tell chatbots about their medical fears, their relationship problems, their beliefs about God and the afterlife.
Advertising built on that archive creates a potential for manipulating users in ways we don’t have the tools to understand, let alone prevent.”
https://globalnews.ca/news/11664538/anthropic-ai-safety-researcher-mrinank-sharma-quits-concerns/
Remember kids, this is why we run our LLMs locally. They definitely have their use, but treat them like a more sophisticated search engine and you'll be fine.
-
“People tell chatbots about their medical fears, their relationship problems, their beliefs about God and the afterlife.
Advertising built on that archive creates a potential for manipulating users in ways we don’t have the tools to understand, let alone prevent.”
https://globalnews.ca/news/11664538/anthropic-ai-safety-researcher-mrinank-sharma-quits-concerns/
@Em0nM4stodon using social media data it was possible to dupe a country into leaving the EU and convincing even more that immigrants are somehow to blame for bad governance.
We should definitely be worried for what’s coming next.
-
“People tell chatbots about their medical fears, their relationship problems, their beliefs about God and the afterlife.
Advertising built on that archive creates a potential for manipulating users in ways we don’t have the tools to understand, let alone prevent.”
https://globalnews.ca/news/11664538/anthropic-ai-safety-researcher-mrinank-sharma-quits-concerns/
@Em0nM4stodon This is about the epidemic of loneliness. So many people are so thoroughly alone that even a chatbot is better than the ache of being alone.
-
“People tell chatbots about their medical fears, their relationship problems, their beliefs about God and the afterlife.
Advertising built on that archive creates a potential for manipulating users in ways we don’t have the tools to understand, let alone prevent.”
https://globalnews.ca/news/11664538/anthropic-ai-safety-researcher-mrinank-sharma-quits-concerns/
@Em0nM4stodon
"I have finished my AI safety research, and found the answer to be NO!" -
“People tell chatbots about their medical fears, their relationship problems, their beliefs about God and the afterlife.
Advertising built on that archive creates a potential for manipulating users in ways we don’t have the tools to understand, let alone prevent.”
https://globalnews.ca/news/11664538/anthropic-ai-safety-researcher-mrinank-sharma-quits-concerns/
@Em0nM4stodon I'm surprised that people are surprised by ads in these things.
its by the playbook.
-
“People tell chatbots about their medical fears, their relationship problems, their beliefs about God and the afterlife.
Advertising built on that archive creates a potential for manipulating users in ways we don’t have the tools to understand, let alone prevent.”
https://globalnews.ca/news/11664538/anthropic-ai-safety-researcher-mrinank-sharma-quits-concerns/
@Em0nM4stodon
Maybe it's time I should flood some AIs with information about #apatheticAgnosticism to counteract the typical #theism. -
“People tell chatbots about their medical fears, their relationship problems, their beliefs about God and the afterlife.
Advertising built on that archive creates a potential for manipulating users in ways we don’t have the tools to understand, let alone prevent.”
https://globalnews.ca/news/11664538/anthropic-ai-safety-researcher-mrinank-sharma-quits-concerns/
@Em0nM4stodon ".. grave concern about the erosion of guiding principles established to preserve the integrity of AI .."
The what now?
-
@Em0nM4stodon why all of those announcements feels like marketing?
@Lioh @Em0nM4stodon can you explain where you read marketing there? I see a clear warning.
-
@Lioh @Em0nM4stodon can you explain where you read marketing there? I see a clear warning.
@pumkin yes, they are all warning but yet selling it. Strange or not? @Em0nM4stodon
-
@pumkin yes, they are all warning but yet selling it. Strange or not? @Em0nM4stodon
@Lioh @Em0nM4stodon I agree with the lobbying we've seen so far from CEOs etc. But the people from the article are leaving the company. Im not surprised by their arguments. But these are the moments, where you can get some more people convinced that there is a lot at stake here (dont want to call them npc's as though they act like some).
-
@Lioh @Em0nM4stodon I agree with the lobbying we've seen so far from CEOs etc. But the people from the article are leaving the company. Im not surprised by their arguments. But these are the moments, where you can get some more people convinced that there is a lot at stake here (dont want to call them npc's as though they act like some).
@pumkin let's just wait and see what this person is going to do next. @Em0nM4stodon
-
@pumkin let's just wait and see what this person is going to do next. @Em0nM4stodon
@Lioh @Em0nM4stodon probably setting up a profile on rentahuman.ai
-
“People tell chatbots about their medical fears, their relationship problems, their beliefs about God and the afterlife.
Advertising built on that archive creates a potential for manipulating users in ways we don’t have the tools to understand, let alone prevent.”
https://globalnews.ca/news/11664538/anthropic-ai-safety-researcher-mrinank-sharma-quits-concerns/
@Em0nM4stodon Our societies are truly astonishing, given that so many obviously incompetent people are able to function day-to-day without it all collapsing.
-
@Em0nM4stodon Our societies are truly astonishing, given that so many obviously incompetent people are able to function day-to-day without it all collapsing.
@edwiebe @Em0nM4stodon
There have always been fools. -
“People tell chatbots about their medical fears, their relationship problems, their beliefs about God and the afterlife.
Advertising built on that archive creates a potential for manipulating users in ways we don’t have the tools to understand, let alone prevent.”
https://globalnews.ca/news/11664538/anthropic-ai-safety-researcher-mrinank-sharma-quits-concerns/
@Em0nM4stodon As a specie, humans appear to be pretty stupid. Lemmings may appear more stupid. But I fail to see others
-
“People tell chatbots about their medical fears, their relationship problems, their beliefs about God and the afterlife.
Advertising built on that archive creates a potential for manipulating users in ways we don’t have the tools to understand, let alone prevent.”
https://globalnews.ca/news/11664538/anthropic-ai-safety-researcher-mrinank-sharma-quits-concerns/
@Em0nM4stodon targeting ads based on search history is one level. Having a fake psychologist that targets ads at you is a whole-nother-level.
-
“People tell chatbots about their medical fears, their relationship problems, their beliefs about God and the afterlife.
Advertising built on that archive creates a potential for manipulating users in ways we don’t have the tools to understand, let alone prevent.”
https://globalnews.ca/news/11664538/anthropic-ai-safety-researcher-mrinank-sharma-quits-concerns/
So, which is more disturbing, the topics people talk about to some computer code, or the fact that we know what those topics are, when the people probably thought they were speaking privately?
I think it's the fact that we know
Also, the concerns re manipulating in the article.
-
N necrosis@chaos.social shared this topic
