#machinelearningfails 검색 결과

Silicon Dales

3 분

Most AIs fail because we ask too much, too soon: silicondales.com/ai/least-to-mo… Least-to-Most Prompting fixes that — breaking problems into steps, guiding models (and people) from independence to mastery. Discover why it outperforms Chain-of-Thought.

SiliconDales's tweet image. Most AIs fail because we ask too much, too soon: silicondales.com/ai/least-to-mo…
Least-to-Most Prompting fixes that — breaking problems into steps, guiding models (and people) from independence to mastery. Discover why it outperforms Chain-of-Thought.

Manu Otel

@manuotel

2 시간

I feel like this is such a bad defeat. Like how's it that bad....??? Tf am I doing wrong to the AI????

DE_ARK

@DE_ARK_Web3

7 시간

🚨 Why ChatGPT Can’t Replace a Crypto Analyst (Yet). AI is powerful… but it still can’t read the markets like you do. 👇

CG

@cgtwtz

. 8. 19.

we are in end game.

Darth Powell

@VladTheInflator

. 8. 18.

Al Jazeera English

@AJEnglish

3 시간

AI models such as ChatGPT routinely misrepresent news events, providing faulty responses to questions almost half the time, a study has found aje.io/jjvnv6

AJEnglish's tweet image. AI models such as ChatGPT routinely misrepresent news events, providing faulty responses to questions almost half the time, a study has found aje.io/jjvnv6

Crémieux

@cremieuxrecueil

2024. 7. 25.

A new paper in Nature found that you cannot, in fact, train AIs on AI-generated data and expect them to continue improving. What happens is actually that the model collapses and ends up producing nonsense.

cremieuxrecueil's tweet image. A new paper in Nature found that you cannot, in fact, train AIs on AI-generated data and expect them to continue improving.

What happens is actually that the model collapses and ends up producing nonsense.

Timothy Caulfield

@CaulfieldTim

2 분

Ouch. AI chatbots fail at accurate news, major study reveals p.dw.com/p/528zp?maca=e… "...four of the most commonly used AI assistants misrepresent news content 45% of the time..."

CaulfieldTim's tweet card. AI chatbots such as ChatGPT and Copilot routinely distort the news and struggle to distinguish facts from opinion. That's according to a major new study from 22 international public broadcasters,...

New study shows widespread AI misinformation

출처: dw.com

Taran Me Up! Taran Me Down!

@tarangoestohell

24 시간

ChatGPT just being blatantly wrong about simple information. I repeat AI is getting dumber at an alarming rate

The Fourth Way

@The4thWayYT

. 8. 19.

we’re not so easily replaceable

QUEEN LIV 🤍

@QueenOlivia99

. 9. 11.

Sometimes when I ask an #AI to do something, it gives me the wrong output. At first, I blame the model. But later I realize… the problem was me. Because the truth is: AI doesn’t fail because the models are weak. It fails because the data it’s given is broken. A thread 🪡

QueenOlivia99's tweet image. Sometimes when I ask an #AI to do something, it gives me the wrong output.

At first, I blame the model. But later I realize… the problem was me.

Because the truth is:
AI doesn’t fail because the models are weak. It fails because the data it’s given is broken.

A thread 🪡

Michael (((Berman)))

@_mberman

2015. 1. 13.

One of these is not like the other #machinelearningfails

Vivek Agarwal

@dr_v_agarwal

2018. 9. 29.

@Twitter (via @Microsoft) detected your tweet as #Haitian creole but then failed to translate it...lol #MachineLearningFails

dr_v_agarwal's tweet image. @Twitter (via @Microsoft) detected your tweet as #Haitian creole but then failed to translate it...lol
#MachineLearningFails

Clint Aveben

@clintaveben

. 5. 9.

🍓𝗧𝗵𝗲 𝗦𝘁𝗿𝗮𝘄𝗯𝗲𝗿𝗿𝘆 𝗣𝗿𝗼𝗯𝗹𝗲𝗺 - 𝘈 𝘊𝘩𝘢𝘵𝘎𝘗𝘛 oopsie.. #AIRealityCheck #MachineLearningFails #PredictiveText #GPT4 #StrawberryGate #AIwithHumans #ArtificialIntelligence #DigitalMarketing #TechWithHumor #ChatGPT #GenerativeAI #TrustButVerify #AIHallucinations

clintaveben's tweet image. 🍓𝗧𝗵𝗲 𝗦𝘁𝗿𝗮𝘄𝗯𝗲𝗿𝗿𝘆 𝗣𝗿𝗼𝗯𝗹𝗲𝗺 - 𝘈 𝘊𝘩𝘢𝘵𝘎𝘗𝘛 oopsie..
#AIRealityCheck #MachineLearningFails #PredictiveText #GPT4 #StrawberryGate #AIwithHumans #ArtificialIntelligence #DigitalMarketing #TechWithHumor #ChatGPT #GenerativeAI #TrustButVerify #AIHallucinations

Dmytro Iakubovskyi

@dima806_dima

. 10. 20.

“Top 5 Machine Learning Mistakes You Should Avoid” by Dima Iakubovskyi medium.com/data-and-beyon…

zed

@zmkzmkz

. 4. 14.

I'm very sorry to disappoint you all, but we did not get the results we wanted. The training loss began to diverge badly compared to baseline at >25B tokens. We could've never known this, since it all looked good when training <5B tokens big thanks to @FAL for the compute though

zmkzmkz's tweet image. I'm very sorry to disappoint you all, but we did not get the results we wanted. The training loss began to diverge badly compared to baseline at &gt;25B tokens. We could've never known this, since it all looked good when training &lt;5B tokens

big thanks to @FAL for the compute though

zed

@zmkzmkz

. 4. 8.

hey guys, I am REALLY desperate for GPUs right now, and I need to scale these results up. I cannot disclose the idea right away, but if anyone here can confirm to lend me at least 8xA100 for ~1-2 weeks we can work together. I promise if these results hold, it will be huge. Lmk 🙏

zmkzmkz's tweet image. hey guys, I am REALLY desperate for GPUs right now, and I need to scale these results up. I cannot disclose the idea right away, but if anyone here can confirm to lend me at least 8xA100 for ~1-2 weeks we can work together. I promise if these results hold, it will be huge. Lmk 🙏

Saidul

@saidul_dev

. 10. 20.

AI isn’t ready for the real world yet. We tested top models on 30 real-robot tasks. The best success rate? 45%. Here’s the reality check.

saidul_dev's tweet image. AI isn’t ready for the real world yet.

We tested top models on 30 real-robot tasks.

The best success rate? 45%.

Here’s the reality check.

Don Johnson

@DonMiami3

. 8. 18.

What stage?

Google Developer Group NYC

@gdg_nyc

2016. 11. 19.

When @random_forests mentioned #MachineLearningFails, room exploded into laughter and things got serious quick! #DevFest16 #DFNYC #Skynet

RV TechLearn

@RVTechLearn

. 12. 19.

"Chihuahua or muffin? 🐶🧁 Sometimes, even AI has a sweet tooth! 😂 #MachineLearningFails #AIHumor #TechieLife #rvtechlearn #letslearninnovateandsucceed"

RVTechLearn's tweet image. "Chihuahua or muffin? 🐶🧁
Sometimes, even AI has a sweet tooth! 😂
#MachineLearningFails #AIHumor #TechieLife #rvtechlearn #letslearninnovateandsucceed"

ASCENTT

@ascentt

. 2. 25.

Discover why the majority of #artificialintelligence fails with this infographic! #MachineLearningFails #TechFailures #AI #AIAnalysis #AIchallenges #AIProblems #AIInsights cc: @Hana_ElSayyed @CurieuxExplorer @mvollmer1 @sallyeaves

ASCENTT

@ascentt

. 2. 18.

RV TechLearn

@RVTechLearn

. 12. 19.

"Chihuahua or muffin? 🐶🧁 Sometimes, even AI has a sweet tooth! 😂 #MachineLearningFails #AIHumor #TechieLife #rvtechlearn #letslearninnovateandsucceed"

EyeWaz

@eye_waz

. 12. 14.

"AI's got its priorities straight: World domination can wait, but this ref's bald head? That's science in action!" 😂😂😂 #AIHumor #MachineLearningFails #TechComedy #AIvsSports #BaldHeadScience #RefCam #TechGoneWild #FutureOfAI #ArtificialIntelligence #FunnyTech

eye_waz's tweet image. "AI's got its priorities straight: World domination can wait, but this ref's bald head? That's science in action!"
😂😂😂

#AIHumor #MachineLearningFails #TechComedy #AIvsSports #BaldHeadScience #RefCam #TechGoneWild #FutureOfAI #ArtificialIntelligence #FunnyTech

ASCENTT

@ascentt

. 12. 9.

Fathiyah

@yarinyagirl1

2024. 10. 6.

AI: I can predict stock markets and win chess tournaments. Also AI: Struggles to tell the difference between a cat and a muffin. 🤷‍♂️🐈🧁 #AILimits #MachineLearningFails #FunnyAI

Ankit Agrawal

@SaintlyPioneer

2024. 6. 29.

Just tried 'teaching' SDXL AI to create my face... and it's hilariously bad! 🤣 It took 1-hour of training to fine-tune base v1 model. Results??? Let's say I won't be using it for profile pics anytime soon!🙃 #SDXL #AIExperiments #MachineLearningFails #0to100xEngineer