I finished reading "Slow River" by Nicola Griffith. An excellent read, 4.5/5. Review below :-) #books #SF #SFF
app.thestorygraph.com/reviews/1d77...
Posts by Emmanuel Dubois
I finished "The Speed of Dark" by Elizabeth Moon. Liked it very much, 4/5 stars. More details below. #readmorebooks #bookreview #sf #sciencefiction
app.thestorygraph.com/reviews/1806...
50,000 downloads!
Glad to see that people are still engaging with the podcast! Thank you all! #podcast #history #achievement
👋 🚀
My month in books. Working on the Nebula project :-)
I just read "Parable of the Talents" by Octavia E. Butler. Remarkable, 5 out of 5. Review below. #OctaviaButler #scifi #readmorebooks.
app.thestorygraph.com/reviews/1621...
AI hierarchy:
Claude : make me a deep analysis of this complex text.
ChatGPT : make a surveillance program for the Pentagon, and don’t ask why.
Gemini : make me this useless and probably copyright infringing illustration.
Mistral : translate this into French
Siri : count down 30min
😅 #AI
The Zone of Thoughts novels are
stronger, but Rainbow’s End is also a good read and touches on aspects of the article :-)
It's fun nostalgia, but the characters are paper-thin. One of the least empathy-demanding novels I can think of. Butler, Le Guin, Vinge... those actually do the work she describes. RPO kind of undermines her own argument :-/
Thanks for the link, good essay. The research on fiction and empathy is solid, the reading gender gap stats are striking, and the point about fathers reading less to sons hit hard.
That said, Ready Player One as an example of politically engaged SF?
To be clear, I’m not saying that well-read, cultured people will automatically behave more wisely. Reading and culture are tools for the mind. Like any tool, their value depends on whether we choose to use them well. #Reading #Fiction #CognitiveScience #CriticalThinking #Books #Literature
Immersing in imaginary worlds isn’t just entertainment. It’s a social simulator that makes us more agile & equipped to understand others. Study: Djikic et al. (2013), “Opening the Closed Mind: The Effect of Exposure to Literature on the Need for Closure.”
Following complex characters and navigating the ambiguity of a story trains our brain to tolerate uncertainty. Unlike a manual, fiction forces us to delay the moment we freeze our judgment. It gives us the time to consider multiple scenarios.
Researchers at the University of Toronto have shown that reading fiction reduces the “need for cognitive closure.” Simply put, it’s that visceral urge to find a definitive, immediate answer to a problem, even if it means jumping to conclusions.
I wondered if this was just a bias from reading too many novels & studying the humanities. So I looked into whether my intuition was grounded in actual research. Turns out, cognitive psychology has looked into exactly this.
Where technical reading gives us facts, fiction gives us perspective and tools to think clearly. The same goes for general knowledge: without it, we remain confined to a rigid way of thinking, unable to put ourselves in other people’s shoes.
I’ve always had the deep conviction that reading quality fiction strengthens the mind & intellectual flexibility. I have nothing against non-fiction or technical books—they provide essential data. But settling for them alone limits our perception. 🧵
Quand j’étais môme, je croyais que “sex appeal” c’était “sexe à piles”. Je ne voyais pas ce qu’ils voulaient dire… et je ne réalisais pas le double sens du tout.
The others can find a ton of data, sometimes more (Gemini), but they are less able to work with it and produce usable results.
The problem wasn’t the models’ “power” or abilities, its design choices. Claude is designed with Constitutional AI and it shows. I thought it was a marketing gimmick but my results confirm this. This produces a system optimised for coherence, synthesis, and respect for user parameters.
But for actual work (not creating idiotic images or asking for encyclopedic level of data), Claude’s consistency, accuracy and ability to synthesize properly was unrivalled. Still, it has to be used as an assistant, an helper, a first draft, or a sorter of data. It can’t replace the person using it.
Mistral tends to hallucinate more than the others. It’s nice that it’s French and open source, but its sorting and analysis capabilities are not on par with Claude. I did find them on par with Chat GPT thought which also hallucinates a lot. Gemini was second best overall and is very feature rich.
Claude did write “I don’t know”, or “this is not 100%” clear on several occasions. None of the others did so, not once. Gemini was able to search more sources than the others, especially in deep search, but it had a tendency to ignore briefs (gave me 8000 words reports when asked for 4000).
Only Claude gave me good, consistent results. The three other seem to suffer from the same basic defect: they can accumulate a ton of data but they aren’t always good at sorting it. And when they fail, they will invent a figure that sounds right instead of writing “I don’t know”.
I’ve been doing a lot of testing over the past several weeks. ChatGPT, Claude, Mistral and Gemini, all in pro modes. Historical research, latest news, deep analysis, long document analysis, establishing patterns, creating html code, sorting data, and many more. My results are as follows:
AI should really be treated as a tool, not as an answer to everything. It can help, it can complement, but it should never replace judgement or skill.
I read “Parable of the Sower” by Octavia E. Butler. 4.5/5, remarkable novel. My review is accessible below. #book #SF #bookreview #dystopian
app.thestorygraph.com/reviews/bfed...
I just finished “A Master of Djinn” by P. Djèli Clark. Quite liked it! 4/5 stars. Review below! #bookreview #SFF #readmorebooks
app.thestorygraph.com/reviews/52e3...
I watched “One Battle After Another”. 3.5/5. Check out my review below. #moviereview #oscars #onebattleafteranother
boxd.it/dtSMEB
Ah, it’s showing up now!