Andi said:

The seahorse emoji phenomenon reveals a curious case of mass misremembrance - there has never been a seahorse emoji, yet both humans and AI language models firmly believe it exists^[1][2].

When asked about the seahorse emoji, large language models respond with complete confidence that it exists, then spiral into confusion when trying to display it, often outputting random fish or horse emojis instead^[2:1]. This behavior stems from the models building an internal “seahorse + emoji” concept that crashes against reality when no matching token exists in their vocabulary^[1:1].

The technical explanation involves the models’ logit lens - as they process the request through their layers, they construct a conceptual blend of “seahorse” and “emoji” that seems perfectly valid until the final output stage, where they’re forced to select the closest available match^[1:2].

Many humans share this false memory, with Reddit threads and social media posts filled with people convinced they’ve seen a seahorse emoji before^[1:3]. While a seahorse emoji was proposed to the Unicode Consortium in 2018, it was rejected and has never actually existed^[2:2].

Adding that it don’t exist in the official Unicode emoji pack, but they do in inofficial packs, eg. here

I never saw this or any ad on YT, using only the Vivaldi trackerblocker and uBO Lite, also the Vivaldi adblocker works fine, but is discovered sometimes by the antii-adblock algorritm from YT. Anyway not possible to avoid the promotions from the Youtuber itself (not confuse with embedded sponsorblocks, which are filtred), also not posible to watch vids “only for members” which are appearing more and more, but I think it’s a legit manner to create incommings for the authors.

ChatGPT - Apertus (FOSS) (Swiss)

OneDrive - Filen (Germany - 10GB free)

Google Photos - vgy.me (UK)

Gdocs/Gdrive/Gmail - Murena Workspace

Google Maps - CoMaps (FOSS- privacy focused)

NSFW isn’t only porn. Use Swisscows as search engine and you’ll understand why it isn’t recommend to block NSFW if you aren’t a young kid.

No problem

Agree with it, despite using an AI search since almost 4 years, Andisearch, because it’s the only one I tested with a pretty good accuracy (~90%). Anyway, it’s always needed to contrast the information in the web, independent of results from AI or not, a lot of BS out there.

To this topic Andi said:

Based on recent examples from 2025-2026, AI search engines frequently provide confident but incorrect answers, demonstrating several key problems:

1. Inconsistent results - Siri with Apple Intelligence gives different wrong answers to the same question when asked multiple times^[1].

2. False confidence - AI provides detailed but completely incorrect information, like Google AI claiming a South Dakota team won North Dakota’s championship^[1:1].

3. Regression in quality - Traditional search results often work better than AI versions. As John Gruber notes, “old Siri… at least recognizes that Siri itself doesn’t know the answer and provides a genuinely helpful response”^[1:2].

4. Poor accuracy even on popular topics - Siri achieved only a 34% accuracy rate when asked about Super Bowl winners, with one stretch of 15 wrong answers in a row^[1:3].

The core issue appears to be that AI search engines prioritize providing definitive-sounding answers over accuracy, making them less reliable than traditional search results that simply link to authoritative sources.

Ads? Ah, those annoying stupid things I watch in TV. Online I don’t see any of these since more than 10 years.

Enough security and privacy with Portmaster, no need of any AI crap in the system.

It won’t

Normal - 4h coding + 2 h debugging

With an LLM - 5 min coding + 24h debugging

There are tons of pages out there with lists of code snippets which can help the devs without the need to use an LLM. It’s naiv to think that LLM which hallucinate in a normal answer won’t do it with coding.

The best weather forecast is always the one from the official weather agency of your country

On Mobile the widget from AEMET, the Spanish Meteorology Agency, on Desktop the inbuild weather widget in Vivaldi (home/new tab page)

Scary the huge amount of services where it is very difficult and even impossible to delete your account. Signed with your blood and your first born as guarantee.

Let work the people for electricity

https://www.pavegen.com/blog/kinetic-pavements

https://www.lboro.ac.uk/news-events/news/2022/april/wearable-material-generates-electricity/

https://spj.science.org/doi/10.34133/2020/7158953

Also confused about that Europe isn’t a country and bigger than the USA.

Worst these cute little guys which you don’t even see

https://www.australiangeographic.com.au/nature-wildlife/2018/04/irukandji-jellyfish-everything-you-need-to-know/

We’ll see on it’s release, but it’s very positive the growing appearance of modular open phones with Linux based OS. Meanwhile there is FairPhone from the Netherland as years ago released mature product, full modular, using LinageOS or e-OS

Good inventions in capitalism has their first use for killing people

Difficult for Wheelchair user, end of the way for pedal Wheelchair users

That don’t make sense, with limited mobility it is more difficult to move a Wheelchair with pedals than walking, maybe with a cane or walker.

There isn’t, the only posibility is creating an own instance of Mastodon or whatever, where you can implement the corresponding rules and filters for kids. The fediverse as such can’t never be safe for kids, due to it’s own structure if safe instances are federated with such which are not family safe. https://blog.elenarossini.com/gotosocial-empowers-you-to-have-your-own-home-on-the-fediverse-with-unique-controls/

Alternatively social media, specific for kids

https://www.internetmatters.org/resources/social-media-networks-made-for-kids/

It is always needed to educate the kids about the risks and for an correct and safe behavior in the network, using safe search engines, eg.SwissCows, private and family safe by design.