Ir para conteúdo

andriy pereplyotkin

Membro da Comunidade
  • Total de itens

    21.441
  • Registo em

  • Última visita

Tudo que andriy pereplyotkin publicou

  1. Tem resultados na carreira mas acho um downgrade. Ainda assim era difícil não ser. Pá, é com cada tamanco que passou por Coimbra fds
  2. Sinto zero isso. Tirando as redes sociais e as páginas de apoio não-oficiais, é um não assunto. Aliás, dado o estado da cidade, acho mesmo que ele ganha porque ela é fraquíssima em debate, por causa da herança Machado e da viragem à direita em todo o país. Não que o PS tenha qualquer quadro alternativo (aliás, até fiquei surpreendido pela positiva com os candidatos a vereador do PS... porque são quase todos independentes). Isto para dizer que não sinto o Zé do Boné num momento de extrema popularidade.
  3. E se não resolver é sinal que ainda não foi o suficiente!
  4. Funerais, casamentos, aniversários, visitas de estado, censura... dá para todas as ocasiões. Que achado!
  5. Por acaso não percebi esse rumor. Que eu saiba eles dão-se bem. Só que o Rui tinha outros projetos para a vida (desportivos e não só) e depois do United decidiu arrancar. Já lá vão quase 10 anos, ia voltar agora? E sim. O Pedro Machado nem isso, dos piores que vi na Académica. E vi muita m*rda.
  6. Acho que estás a ser simpático, provavelmente era só um ministério e o resto tudo corrido a secretaria. Para se perceber bem quem é que manda. Um Ministério da Raça e do Ultramar ficava muito bem.
  7. Get ready to learn salvadorean buddy
  8. Saudades Eva Carneiro. Eu acho é que são bocas para o Mário Branco.
  9. É assim, por mim tinha ficado, mas também acho que não é jogador para mais que uma Liga 3. Pelo menos para uma equipa com ambições. De qualquer das formas, a estratégia será mesmo a curto prazo. É um tipo com 27 anos que tem uma época numa liga profissional. Ofereceram-lhe 5k/mês. É hora de fazer um pé-de-meia.
  10. Por mim tinha ficado em Coimbra, mas disseram-me que está a receber (bem) mais do que qualquer equipa da Liga 3 ofereceu. E não faltaram convites.
  11. https://www.pimlicojournal.co.uk/p/mps-are-almost-certainly-using-chatgpt Clearly, the ‘rule of 3’ and the ‘not just, but’ formula are also used by humans; if they were not, they would not be used by ChatGPT so often. However, in my view, you would struggle to find any text written by a human that uses them nearly as much as in this speech. That is before even considering some of the other suspicious elements in the speech. In fact, I left out highlighting some of the melodramatic short statements (‘We need to call it out for what it is’ or ‘The moral imperative could not be clearer’) which are, in my view, usually strongly suggestive of AI, but may have another cause in this context: Anthony Powell noted that some officers altered their speech during the Second World War to take on a more Churchillian meter; it is entirely possible that Morton has done a similar thing, but with the much less inspirational figure of Boris Johnson. There are also some other incongruities and strange phrases that I have not flagged because they could just be put down to bad writing. This is not a piece that is intended to single out Morton. If our supposition that this speech was substantially written by ChatGPT is indeed correct, our data shows that she is not alone. There are therefore plenty of other examples of suspicious speeches, and I will briefly quote short excerpts from two others with my comments. I did not search particularly far or wide to find these speeches. They were selected more or less at random, and there may be far worse cases that I’ve missed. I hope they also show how it can be difficult to find and quantify the amount of ChatGPT use: ‘tears are a natural result of such a devastating awareness’ — a phrase used in the next speech excerpt — sounds very weird, but there is no reason it would show up in any statistical analysis. First, Shockat Adam, the newly-elected Independent MP for Leicester South, speaking in a debate about long-term medical conditions: Second, Mohammad Yasin, the Labour MP for Bedford since 2017, speaking in a debate about university funding. This speech is somewhat less suspicious than the previous two, but we can still flag a number of phrases even in this short passage: To repeat again: there is no definitive proof that any of these three speeches made use of AI; however, there is at least some reason for suspicion. By contrast, this speech by Keir Starmer in PMQs is a good point of comparison as a speech that has far fewer of these verbal tics that are closely associated with AI-generated text. If I am correct in suspecting the widespread use of AI to generate Commons speeches, then it is profoundly depressing that some of our MPs (or at least the people working for them) appear to be so unimaginative, so illiterate, so empty, that they are willing to outsource what should be an MP’s greatest task and honour to a machine. I do not want a future where boring ex-public sector workers read out AI-generated statements about their hobby horses in a sterile, semi-circularly arranged ‘Parliament’ with enforced silence. The art of the parliamentary speech and the tradition of confrontational debate are worth promoting and protecting. Appendices Appendix A Many of the terms reported to be favoured by AI are: Words that have been heavily used in certain other contexts within Parliament, such as ‘seamless’ or ‘framework’ during Brexit negotiations, ‘landscape’ when talking about construction, or ‘realm’ in ‘defence of the realm’. Therefore some of these words have been left out, although a couple (landscape, ecosystem) are included for reference. Words or phrases that are overused when ChatGPT is used in the majority of cases — academic work, reports, and emails — but not necessarily so if an MP is prompting it to ‘write me a speech to read in Parliament about [X]’. ‘Delve’ might be a case of this. Simply not used frequently enough to be able to draw any inferences from the data. Appendix B Annual adjustment factors: 2007 (1.0236), 2008 (0.9819), 2009 (1.0988), 2010 (1.0299), 2011 (0.9772), 2012 (1.0675), 2013 (1.0130), 2014 (1.0548), 2015 (0.9999), 2016 (0.8190), 2017 (1.0207), 2018 (0.7943), 2019 (0.9556), 2020 (1.0791), 2021 (0.9729), 2022 (0.9816), 2023 (1.1613), 2024 (1.2006), 2025 (1.2872). Appendix C Of the four words, ‘delve’, ‘intricate’, and ‘elevate’ are used at relatively low frequencies, which makes the data noisier and it harder for a pattern to be visible. All three of those do actually have a positive Z-score in 2024 and 2025 — except ‘delve’ in 2025, which is ≈0 — i.e., they are being used more frequently than average; it’s just not at a statistically significant level or part of a meaningful pattern.
  12. Todas as eleições deviam ter um idiota destes só para haver termo comparativo.
  13. Estás calvo de saber que o homem tinha cabelo.
  14. Vá, totalmente culpa do clube não é. É culpa da autarquia e das forças de segurança locais, que não identificaram a ocasião atempadamente. A notícia que o @Diogo_CFB publicou explica bem o assunto. O Arouca clube também é vítima.
  15. Ou os dois laterais por dentro e o Neto e o João por fora. Eu sei lá.
  16. Em voos de fim-de-semana, naturalmente mais cheios. O Nacional tem obviamente argumentos. Num caso destes a Liga tem de tentar encontrar uma data que satisfaça as duas partes. Esta conversa não é digna para ninguém.
×
×
  • Criar Novo...