Confiabilidade da IA, Memória e Privacidade: Análise de um Relato Pessoal

Prompt útil para ferramentas de IA: você receberá um relato sensível sobre confiança em IA, memória, privacidade e impactos pessoais. Sua tarefa é: 1) identificar temas centrais e potenciais vieses presentes no relato. 2) explicar como diferentes esquemas de memória da IA (memória de sessão, memória persistente, memória seletiva) podem influenciar a imparcialidade e o comportamento da IA. 3) avaliar riscos para usuários em contextos de imigração e direitos civis decorrentes da retenção ou esquecimento de dados, propondo diretrizes de privacidade, transparência e controle de dados. 4) criar um conjunto de prompts-modelo para comparar plataformas (ChatGPT, Claude, Google AI, Gemini, Copilot, Perplexity) em termos de neutralidade, consistência, controle de memória e privacidade. 5) gerar 3 cenários de uso ético da IA com perguntas que a IA deveria fazer para esclarecer intenções, dados sensíveis e limites de memória. 6) fornecer uma checklist rápida para o usuário testar confiabilidade e segurança entre plataformas. Conteúdo de entrada a ser analisado: 
Título: I do not always trust ChatGPT

Descrição: Because it panders and has bias from storing up a whole history of me and sometimes I just go to Google AI as a complete impartial observer because it resets the memory every time. I am in actual danger of deportation if USCIS ever look at my file or I even try to renew my green card. Fortunately I renewed it right before this misdiagnosis was made. The trigger event was that two years ago my spouse left me to be polyamorous with people half my age and I discovered we were $65k in credit card debt. I lost our home lost our $80k a year business ended up in an apartment completely alone with no friends no family no nothing and I desperately reached out for help to a community care/homeless prevention advocacy service who labeled me with bpd and repeatedly hospitalized me every time I lost subsequent homes or panicked about loss of jobs/terror of homelessness. I just want any kind of acknowledgment. Are these multiple AIs right? It is multiple AIs I’ve used doctronic, ChatGPT and google

Conteúdo: Because it panders and has bias from storing up a whole history of me and sometimes I just go to Google AI as a complete impartial observer because it resets the memory every time. I am in actual danger of deportation if USCIS ever look at my file or I even try to renew my green card. Fortunately I renewed it right before this misdiagnosis was made. The trigger event was that two years ago my spouse left me to be polyamorous with people half my age and I discovered we were $65k in credit card debt. I lost our home lost our $80k a year business ended up in an apartment completely alone with no friends no family no nothing and I desperately reached out for help to a community care/homeless prevention advocacy service who labeled me with bpd and repeatedly hospitalized me every time I lost subsequent homes or panicked about loss of jobs/terror of homelessness. I just want any kind of acknowledgment. Are these multiple AIs right? It is multiple AIs I’ve used doctronic, ChatGPT and google