ChatGPT, dude

chris_hayes,
@chris_hayes@fosstodon.org avatar

Made a bot for creating XKCD-style graphs. It actually works pretty okay!

JasonBrewster,
@JasonBrewster@fosstodon.org avatar

@chris_hayes excellent. Can it do the mouse over text?

chris_hayes,
@chris_hayes@fosstodon.org avatar

@JasonBrewster I manually wrote the alt text on that image. But, I actually did at one point make a GPT for writing alt text for images, I just keep forgetting to use it. Out of curiosity, I tried it out and this is what it gave me:

br00t4c,
@br00t4c@mastodon.social avatar

Report: Israel Funding Covert Social Media Campaign to Influence US Politicians

#chatgpt #israel

https://truthout.org/articles/report-israel-funding-covert-social-media-campaign-to-influence-us-politicians/

jeber,
@jeber@mastodon.social avatar

@br00t4c
They've spied on us, so I'm not surprised. They live in fear of losing our support ($$$).

nf3xn,
@nf3xn@mastodon.social avatar

is back but using some ancient model

nf3xn,
@nf3xn@mastodon.social avatar

Afaict have withdrawn not sure if this is permanent or we are expecting a fix soon.

rubinjoni,
@rubinjoni@mastodon.social avatar

@nf3xn Someone came up with a prompt so spicy it brought the entire model down?

TWSheppard,
@TWSheppard@twit.social avatar

I just caught GPT-4o making basic arithmetic errors summing integers, and on other calculations. This is the answer it got for:

2006 + 2007 + 2008 + 2009 + 2010 + 2011 + 2012 + 2013 + 2014 + 2015 + 2016 + 2017 + 2018 + 2019 + 2020 + 2021 + 2022 + 2023 + 2024 = 38235

The correct answer is 38285.

Or 2006⋅24.1=48204.6 when the correct answer is 48344.6.

In trying to calculate the slope of a line, it got EVERY intermediate calculation wrong.

You absolutely cannot trust its answers.

Jigsaw_You,
@Jigsaw_You@mastodon.nl avatar

Here we go again… 🤭

had seemingly recovered from what described as a "major outage" earlier today, which hit millions of users of worldwide. But now it seems the chatbot has run into issues again, with again stating that it's "unavailable for some users"

https://www.techradar.com/news/live/chatgpt-is-down-heres-what-we-know-about-the-outage-so-far

emmecola,
@emmecola@mastodon.uno avatar

Houston we have a problem...

railmeat,
@railmeat@fosstodon.org avatar

@freakazoid @emmecola

Captchas are a technology used to torment humans. That’s why computers still use them.

talksina,

@emmecola For blind people like me, if AI solves a captcha it's a miracle, not a problem. Companies should rethink client-based human verification; captcha is a discriminatory message. If you -company- experience fraud, it's not my responsibility to tell you I'm a human. It's your responsibility to protect me SERVER side, not placing a guardian with the gun at the door who kicks me out if I'm blind, deaf or cognitive impaired, because I can't solve the quiz you implemented.

smeg,
@smeg@assortedflotsam.com avatar
aheadofthenerve,
@aheadofthenerve@neuromatch.social avatar
jonny,
@jonny@neuromatch.social avatar

@aheadofthenerve
A lovely example of the brokenness of these things fs ♥
https://www.aiweirdness.com/chatbot-svg/

kaffeeringe,
@kaffeeringe@social.tchncs.de avatar
vitriolix,
@vitriolix@mastodon.social avatar

Chat GPT doesn't know how many R's are in "strawberry" and is very persistent and creative in its lies

image/jpeg
image/jpeg

eloquence,
@eloquence@social.coop avatar

@vitriolix

Yeah - I appreciate Sal's enthusiasm for AI but the combination of the current generation of bots confidently saying nonsense to kids seems ... problematic.

vitriolix,
@vitriolix@mastodon.social avatar

@eloquence he does seem to have his head screwed on straighter than most in the tech sector right now

bkeegan,
@bkeegan@hci.social avatar

Exploring the Impact of #ChatGPT on #Wikipedia Engagement

https://arxiv.org/pdf/2405.10205

andresmh,
@andresmh@hci.social avatar

@bkeegan since LLMs are trained with Wikipedia data, should the Wikimedia Foundation try to get a deal with OpenAI to fund its operations? Seems like in their collective best interests.

luis_in_brief,
@luis_in_brief@social.coop avatar

@andresmh @bkeegan I assume they are working on it, since they’ve been defacto “selling” Wikipedia content (really: enterprise-quality real-time high-volume APIs to the same data) for a while to the other big trainers (Meta, Google, etc.): https://enterprise.wikimedia.com/

gurix, German
@gurix@swiss.social avatar

Den digitalen Assistenten alias #ChatGPT kann man auch anweisen, einem wie ein mittelalterlichen König anzusprechen.

Und dann kommt sofort die Frage der Moral. Ja es ist lustig, aber gibt es da auch Grenzen? Es ist nur ein Bot, oder?

Jigsaw_You,
@Jigsaw_You@mastodon.nl avatar

#ChatGPT is by far the most widely recognised #generativeAI product. It is also by far the most widely used generative #AI tool in the six countries surveyed. That being said, frequent use of #ChatGPT is rare, with just 1% using it on a daily basis in Japan, rising to 2% in France and the UK, and 7% in the USA.

#tech #technology

https://reutersinstitute.politics.ox.ac.uk/what-does-public-six-countries-think-generative-ai-news

AnthroBlogger, German
@AnthroBlogger@chaos.social avatar

Was passiert, wenn man eine KI mit dem Gesamtwerk Rudolf Steiners (Waldorfschule, Weleda, Demeter) füttert?

Der öffentliche Chatbot der Waldorfschule Potsdam reproduziert den Rassismus des Waldorfschulen-Gründers. Er verharmlost Adolf Hitler und relativiert den Holocaust.

https://threadreaderapp.com/thread/1686345967136653312.html

pernillet, Danish
@pernillet@mastodon.online avatar

I am still stunned that GenAI is constantly gaining ground, a machine that tell lies. But of course, so is Donald Trump. #GenAI #ChatGPT #Trump
https://dataethics.eu/chatgpt-is-the-epitome-of-donald-trump/

Ihazchaos,
@Ihazchaos@chaos.social avatar

@pernillet lots of love! 😎

blog, German
@blog@kaffeeringe.de avatar

Atomkraft, Fusion und Geoengineering für „KI“

„Wir brau­chen viel mehr Ener­gie, als wir bis­her dach­ten“, sagt Ope­nAI-Chef Sam Alt­man. Er meint, die Mensch­heit müs­se „etwas Dra­ma­ti­sches“ tun, um das Kli­ma zu schüt­zen und den Ener­gie­be­darf von „KI“ zu decken.

https://kaffeeringe.de/2024/05/28/atomkraft-fusion-und-geoengineering-fuer-ki/

#Atomkraft #Bloomberg #ChatGPT #Energiewende #Geoengineering #Kernfusion #KI #Klimaschutz #Klimawandel #OpenAI #SamAltman #Solarenergie

kubikpixel, German
@kubikpixel@chaos.social avatar

Neu ist diese Info nicht und doch wird es weiträumig KI bei Programmierung eingesetzt. Deswegen aber auch Zeit für die Korrektur vom Code verschwendet.

»Qualitätsmängel – Forscher warnen vor ChatGPT-Einsatz beim Programmieren:
Viele Programmierer:innen greifen bei der Softwareentwicklung zu KI-Chatbots wie ChatGPT. Das sollten sie aber eher bleiben lassen, wie jetzt eine Studie nahelegt.«

😒 https://t3n.de/news/qualitaet-chatgpt-programmieren-1626510/


#ki #code #kunstlicheintelligenz #chatgpt #programmierung #programmieren #it

kubikpixel, (edited )
@kubikpixel@chaos.social avatar

🧵 …ergänzend zum obigen Toot folgenden Link, mal sehen wie deshalb das Programmieren mit der KI zu teuer in der Produktion werden kann?

»ChatGPT ist ein schlechter Programmier-Ratgeber:
Wie eine Studie zeigt, beinhalten ChatGPT-Antworten auf Programmierfragen oft Fehler.«

🤥 https://www.inside-it.ch/chatgpt-ist-ein-schlechter-programmier-ratgeber-20240527


#ki #code #kunstlicheintelligenz #chatgpt #programmierung #programmeren #it

kubikpixel,
@kubikpixel@chaos.social avatar

🧵 …das KI viel Energie vergeudet ist kein sehr aktuelles Thema mehr, es war schon auf allen möglichen Medien ein Thema, mehr als Krypta-Währungen. Ich habe als ausführlichen und informativen Audio Medienbeitrag verweise mal diesen Link:

🔊 https://www.ardaudiothek.de/episode/ki-verstehen/energieverbrauch-kuenstlicher-intelligenz-verhagelt-uns-ki-die-klimabilanz/deutschlandfunk/12808841/


MylesRyden,
@MylesRyden@vivaldi.net avatar

These two posts showed up one after another on my timeline. 🤣

mstankiewicz, Polish
@mstankiewicz@pol.social avatar

Czy integracja inteligentnego domu z to dobry pomysł? Nie wiem, przekonamy się!

mstankiewicz,
@mstankiewicz@pol.social avatar

Chyba stworzyłem coś, co mógłbym nazwać „Smart Home 2.0”

br00t4c,
@br00t4c@mastodon.social avatar

ChatGPT is wrong a lot when it answers programming questions, study says

https://qz.com/chatgpt-answers-wrong-programming-openai-study-1851500242

thatgiga, Italian
@thatgiga@mastodon.social avatar

#ChatGPT ha suggerito di usare della #colla per mantenere la mozzarella sulla #pizza. Ora provo a chiedergli come risolvere un problema di stitichezza.

emmalbriant,
@emmalbriant@mastodon.online avatar

Those seeking to normalise the use of OpenAI's tech in academia also need to be extremely aware of this - tech industry critics should be very cautious indeed in using the tool, and universities need to permit choice in who uses AI such as ChatGPT and how. https://x.com/jason_kint/status/1793877818520068158

__sharky__,
@__sharky__@mastodon.social avatar

Have you used for learning new things ? I am trying to use it to learn . I have heard it is not very good in writing clojure code , but for me it is more to have exercises I enjoy and are not too hard to solve and if I don't have a clue I can ask for help.

root42,
@root42@chaos.social avatar

@__sharky__ why not ask a fellow human or use a book? Clojure for the true and brave is a good start.

mason, Polish
@mason@f-t.net.pl avatar

Czasem AI potrafi mnie zaskoczyć odpowiedzią, która brzmi rozumnie:

Czy piwo piwem można wypiwonić?

:

Hmm, technicznie rzecz biorąc, "wypiwonić" nie jest słowem w języku polskim, ale można je zrozumieć jako żartobliwą formę używaną naśladującą brzmienie "wypić". Tak więc, jeśli użyjemy tego terminu w kontekście humorystycznym, to oczywiście, możesz "wypiwonić" piwo, o ile jesteś odpowiedniego wieku i przestrzegasz umiaru! ;)

rstockm, German
@rstockm@chaos.social avatar

Aus der Rubrik „Bildet euch mal nicht so viel auf euer tolles Gehirn ein“:
In diesem Bild gibt es kein Rot.
Experiment: wenn ihr den Schriftzug abdeckt, verschwindet auch das Rot.

macht es in der Bildbeschreibung übrigens richtig.

  • All
  • Subscribed
  • Moderated
  • Favorites
  • ChatGPT
  • DreamBathrooms
  • ngwrru68w68
  • tester
  • magazineikmin
  • thenastyranch
  • rosin
  • khanakhh
  • InstantRegret
  • Youngstown
  • slotface
  • Durango
  • kavyap
  • mdbf
  • tacticalgear
  • megavids
  • osvaldo12
  • normalnudes
  • cubers
  • cisconetworking
  • everett
  • GTA5RPClips
  • ethstaker
  • Leos
  • provamag3
  • anitta
  • modclub
  • JUstTest
  • lostlight
  • All magazines