meltedcheese (@meltedcheese@c.im)
@robotistry That is a fascinating and counterintuitive result. Thanks for the reference.
When you think about it, many people around the world have some degree of belief in animism, that all objects may contain spirits; in many places it is a key part of religious practice. Robots are no different.
The danger is over-reliance on AI/robots based on trust that is not justified. People have innate means of assessing trust, including those related to social interaction. That includes non-verbal communication. Robots will become more and more intelligent, communicate with us in natural manner and language, seem “alive”, and resemble us. This will naturally increase trust and over-reliance unless the information we get represent “true signals.” Today, there is no way to assure this even by inspection.
The least worst case is human disappointment with machine performance. The near worst cases are failures in life critical and mission critical systems. I know plenty of examples. The worse worst case is if robots “cheat” to achieve goals or otherwise deceive us (and other intelligent systems) into believing they can do things they can’t, or do it better, or know more. We have seen these behaviors emerge in numerous lab examples and now we all have seen it in the #LLMs that are deployed such as #ChatGPT. No programming by villains required! This is why I am opposed to creating robots that are DELIBERATELY intended to manipulate people into believing something false — such as “caring” — the root of this thread.
#AI #robotics #trust #emotion #psychology #ethics
Show Original Post
p450 (@p450@mastodon.social)
De humanos para humanos: rascunho do poema Mandrágora.
Porque arte dá uma trabalheira sim. E geralmente começa com bagunça.
Foto de jul. 2021.
https://paulo57v.medium.com/mandragora-6caf0517d006
#NoAI #SemIA #poema #livros #ChatGPT
Show Original Post
teknofixitsolutions (@teknofixitsolutions@mstdn.social)
OpenAI is finally rolling out a toggle that allows you to decide how hard the GPT-5-thinking model can think. This feature is rolling out to Plus and Pro subscribers.
#Tech #InfoSec #Microsoft #ChatGPT #ArtificialIntelligence #OpenAI #Computing #TechNews #Blog #Technology #News #Travel #Business #Finance #USA #EU #Asia #Africa #SocialMedia #Mastodon
Show Original Post
tugatech (@tugatech@masto.pt)
YouTube aponta o dedo aos ad blockers pela quebra nas visualizações
🔗 https://tugatech.com.pt/t71957-youtube-aponta-o-dedo-aos-ad-blockers-pela-quebra-nas-visualizacoes
#chatbots #ChatGPT #google #ia #servidor #youtube
Show Original Post
notes (@notes@mk.phreedom.club)
#Jan.
#Chat with #AI
without #privacy concerns.
https://jan.ai
https://github.com/menloresearch/jan
#OpenSource #ChatGPT #Linux #Windows
Show Original Post
habr (@habr@zhub.link)
Руководство по ChatGPT: правильно выбираем модель и режим рассуждений
Когда OpenAI только готовила к выпуску GPT-5, то одной из фишек этой модели назывался "роутер" — специальный режим, который будет оценивать сложность запроса пользователя и передавать его наиболее подходящей модели. К сожалению, в реальности все пошло наперекосяк и сейчас в ChatGPT даже больше настроек, чем было до выхода GPT-5. Давайте в них разберемся.
https://habr.com/ru/articles/949144/
Show Original Post
KOKEdit (@KOKEdit@mastodon.social)
From Computerworld: “OpenAI, the creator of #ChatGPT, acknowledged in its own #research that #large #language #models will always produce #hallucinations due to #fundamental #mathematical #constraints that cannot be solved through better engineering.” https://www.computerworld.com/article/4059383/openai-admits-ai-hallucinations-are-mathematically-inevitable-not-just-engineering-flaws.html
Show Original Post
danish_akhtar7 (@danish_akhtar7@mastodon.social)
The only reason that I don't use ChatGPT is that I already have Too Many Online Accounts. I don't want to log-in for yet another online service.
I'm happy with Gemini and Copilot, even if they are less capable than ChatGPT.
#mastodon #fediverse #opinion #quote #tech #AI #thought #insights #discover #explore #analysis #writing #commentary #author #journal #WritingCommunity #technology #app #software #gemini #ChatGPT #Copoilet #online
Show Original Post
me (@me@social.maru.my)
Bisa dibilang seharian sibuk reverse enginer prompt buat jadi customGPT, bisa bikin infografis atau poster sendiri kek gini. #AI #chatgpt
Show Original Post
kagan (@kagan@wandering.shop)
AI Bros: "It's getting better all the time, soon it won't make mistakes, really!"
OpenAI, makers of ChatGPT: We just proved mathematically that "hallucinations" are inevitable.
#AI #AICrap #AiResearch #GenerativeAI #OpenAI #ChatGPT
Show Original Post
Okuna (@Okuna@social.tchncs.de)
Interesting study (thanks @jeffjarvis ) on #ChatGPT and #Google #Search.
Links when you search for content:
„… in the United States, where Google results leaned more toward Brand content at 32.9 percent and Social at 15.4 percent, while AI search emphasized Earned content at 92.1 percent with negligible Social references“
„The automotive category produced one of the clearest
contrasts. In Canada, Google returned 36.6 percent Brand, 22.8 per-
cent Social, and 40.6 percent Earned content. AI search, however,
offered 69.1 percent Earned, 30.9 percent Brand, and no Social re-
sults. In the United States, Google leaned 39.5 percent Brand, 15.4
percent Social, and 45.1 percent Earned, while AI search delivered
81.9 percent Earned and 18.1 percent Brand.“
https://arxiv.org/pdf/2509.08919
Show Original Post
2025 (@2025@rbfirehose.com)
Poynter: Is your mayor using ChatGPT? Here’s how to FOIA around and find out. “The stories highlight the need for AI literacy as the technology embeds deeper into our lives, even in ways we may never see. So, I reached out to [Nate] Sanford in an email conversation to find out why and how he used the Freedom of Information Act to obtain ChatGPT logs, and what they say about where we’re […]
Show Original Post