#negativeresults

10/

• Surprisingly often associated: a refusal "to disclose their #NegativeResults or reveal the process towards obtaining their solutions", to reuse the words in [B] in another context

• #NegativeResults sometimes take the form of clear conceptual mistakes or serious #ScientificCode errors, easy for experts to "smell" even if undisclosed and "secretly" fixed (later on)

• The "refusal to disclose" sometimes appears to be unconscious, as if the ability to understand its gravity were lacking

13/ Blurred spectrum between two refusal ways to disclose #NegativeResults and the process towards obtaining any given "successful" solution

• #generativeAI companies using these "opaque" solutions as advertising and with goals "severely misaligned" [3] vs. the goals of the #ScientificCommunity

• #scientific/ #technical experts increasingly adopting an "opaque" process to deliver "successful" solutions, hiding (or simply being more and more unable to keep up with) even the more severe pitfalls

4/ in [B] @tao@mathstodon.xyz also notes how "there are no clear frontiers that are separating the "«[ #genAI]-feasible» problems from the «[ #genAI]-hard» problems (which certainly still exist, given that the difficulty level of problems are unbounded, and can even be undecidable). This is [...] also compounded by the refusal of [ #genAI] companies to disclose their #NegativeResults or reveal the process towards obtaining their solutions" which brings to mind #CherryPicking and other biases in #science [1]

7/ On the omission and invisibility of #NegativeResults, [1] cites "an elegant, comic paper" [2] showing how even a most improbable thesis can be defended this way: "how to craft analyses to prove listening to the Beatles made undergraduates younger" [1] Which kind of "meta"-thesis on #genAI itself might be defended by actors investing "enormous amount of computational resources" with "the refusal [...] to disclose their #NegativeResults or reveal the process towards obtaining their solutions"?