Ahmed Fessi
LightNews LightNews
banner
ahmedfe.bsky.social
Ahmed Fessi
@ahmedfe.bsky.social
19 followers 37 following 160 posts
Posts Replies Media Videos
ahmedfe.bsky.social
Mistral 3: smaller footprints and faster inference are what keep real systems stable and affordable.
#AIEngineering #Data #LLMOps #Mistral
Introducing Mistral 3 | Mistral AI
A family of frontier open-source multimodal models
mistral.ai
December 16, 2025 at 7:35 AM
ahmedfe.bsky.social
Anthropic’s “Interviewer” hints at a future where evaluation tools are precise, transparent, and easier to trust.
#AI #DataScience #AIEthics #TechInsights
www.anthropic.com/research/ant...
December 15, 2025 at 3:31 PM
ahmedfe.bsky.social
Efficiency in AI comes from removing friction, not just adding power.
#MachineLearning #DigitalTransformation #AI
developer.nvidia.com/blog/focus-o...
December 15, 2025 at 8:49 AM
ahmedfe.bsky.social
AI can sound certain while being wrong. A nice approach to add more assertion and less probability
#AI #Data #ResponsibleAI #AIGovernance #DigitalTransformation
steerlabs.substack.com/p/confident-...
December 14, 2025 at 2:56 PM
ahmedfe.bsky.social
Qwen3 Omni Flash is a reminder that speed, multimodality, and efficiency matter more than raw model size. #AI #LLM #MultimodalAI #DigitalTransformation
qwen.ai/blog?id=qwen...
December 14, 2025 at 7:51 AM
ahmedfe.bsky.social
How will OpenAI monetize ChatGPT?
Well, easy: Ads.

This screenshots is from ChatGPT Android App code and shows it contains features to manage Ads.

#ChatGPT #Ads
December 1, 2025 at 9:40 PM
ahmedfe.bsky.social
Claude AI helped complete tasks 80% faster, changing how we work and spend our time.
#AI #Productivity #ClaudeAI #Data #DigitalTransformation
www.anthropic.com/research/est...
November 28, 2025 at 10:04 AM
ahmedfe.bsky.social
Claude Opus 4.5 handles long conversations well but still needs careful human review.
#AI #DigitalTransformation #Data
www.anthropic.com/news/claude-...
November 27, 2025 at 1:08 PM
ahmedfe.bsky.social
IT project failures: Misaligned goals, unclear requirements, and poor planning matter far more than tools.
#DigitalTransformation #AI #Data
spectrum.ieee.org/it-managemen...
November 27, 2025 at 6:57 AM
ahmedfe.bsky.social
When things work well. Seems to me like a good encryption!
www.bbc.com/news/article...

#Encryption #LockDown #Decryption #Cantdecrypt #PQC
Cryptology firm cancels elections after losing encryption key
The International Association for Cryptologic Research - created to study secure communication - said it was an "honest human mistake."
www.bbc.com
November 26, 2025 at 4:45 PM
ahmedfe.bsky.social
LLM extension: Structured reasoning can beat raw power when models get room to think.
#AI #LLM #MachineLearning #DataScience #AIGovernance #AIFuturehttps
//www.sawyerhood.com/blog/llm-extension
November 25, 2025 at 5:28 PM
ahmedfe.bsky.social
Want to learn about #SustainableAI ?
Get a free access to my new course that I just launched.
All you need to do is to fill this form, and you will receive in a couple of days a free coupon! Ensure you Like the post!

forms.gle/UGMtwvJkExTo...

#AI #Learning #LearnAI #Course #NewCourse #Udemy
Sustainable AI - Giveaway
Are you looking to learn more about Sustainable AI : Understanding AI Environmental Impact and what to do about it? With the launch of my new course on Sustainable AI: I am offering, 100 free coupo...
forms.gle
November 25, 2025 at 2:39 PM
ahmedfe.bsky.social
OpenAI’s latest update is a reminder that AI real progress means building safety and transparency into the system itself.

#AI #AIGovernance #OpenAI #AIAlignment #ResponsibleAI
openai.com/index/ai-pro...
November 25, 2025 at 7:35 AM
ahmedfe.bsky.social
File Search in Gemini API automates embeddings and storage so you can focus on delivery instead of plumbing.
#AI #Analytics #EnterpriseData
blog.google/technology/d...
November 24, 2025 at 3:31 PM
ahmedfe.bsky.social
Useful AI is the kind that helps people decide faster and work smarter. The human layer remains the anchor, even as the tools evolve. And a nice discussion on the bubble aspect
#AI #TechLeadership #DigitalTransformation
www.chrbutler.com/what-ai-is-r...
November 24, 2025 at 8:49 AM
ahmedfe.bsky.social

#DigitalTransformation #RustLang #OpenSource
github.com/longbridge/g...
November 16, 2025 at 2:56 PM
ahmedfe.bsky.social
Treating text like visuals changes embeddings. Interesting and counter intuitive
#AI #MachineLearning #Innovation
www.seangoedecke.com/text-tokens-...
November 16, 2025 at 7:51 AM
ahmedfe.bsky.social
We’re in AI’s dial-up phase messy, fascinating, and full of small breakthroughs that shape real habits.

#AI #TechTrends #DigitalTransformation #Innovation #MachineLearning
www.wreflection.com/p/ai-dial-up...
November 15, 2025 at 1:35 PM
ahmedfe.bsky.social
When verification takes hours and creation takes seconds, trust becomes the real casualty. #AI #CyberSecurity #OpenSource
devansh.bearblog.dev/ai-slop/
November 15, 2025 at 7:01 AM
ahmedfe.bsky.social
OpenAI drew the line clearly. AI can assist, not diagnose or defend.

#AI #Ethics #DigitalTransformation #ResponsibleAI #OpenAI
www.ctvnews.ca/sci-tech/art...
ChatGPT users can’t use service for tailored legal and medical advice, OpenAI says
OpenAI dispelled suggestions that it’s changing its terms around legal and medical advice.
www.ctvnews.ca
November 14, 2025 at 6:07 PM
ahmedfe.bsky.social
AI is filling the internet fast, but trust fades when meaning gets lost. Quality and human intent still matter most.

#AI #Data #DigitalTransformation #Trust #AIEthics
arnon.dk/the-trust-co...
November 14, 2025 at 10:04 AM
ahmedfe.bsky.social
Tried Data Formulator smart concept turning text into analysis, still early but genuinely useful. Seems like more just a wrapper. Data Privacy topic to check also.

#AI #DataAnalytics #DigitalTransformation
data-formulator.ai
November 13, 2025 at 1:08 PM
ahmedfe.bsky.social
AI image models shine differently pick the one that fits your workflow.

#AI #ImageGeneration #MachineLearning #Data #DigitalTransformation #PracticalAI
latenitesoft.com/blog/evaluat...
We ran over 600 image generations to compare AI models - LateNiteSoft Blog
We benchmarked OpenAI gpt-image-1, Google nanoBanana, and Seedream across 600+ real-world photo edits, see latency, cost, and quality by task
latenitesoft.com
November 13, 2025 at 6:57 AM
ahmedfe.bsky.social
Open source keeps AI and data flowing, yet most of it runs on unpaid effort.
#AI #Data #OpenSource #DigitalTransformation #TechEthics
thenewstack.io/ffmpeg-to-go...
FFmpeg to Google: Fund Us or Stop Sending Bugs
A lively discussion about open source, security, and who pays the bills has erupted on Twitter.
thenewstack.io
November 12, 2025 at 5:06 PM
ahmedfe.bsky.social
Nested Learning lets models adapt without forgetting, saving retraining time, but adds complexity and risk of bias propagation. #AI #MachineLearning #ContinualLearning #DataScience #DigitalTransformation
research.google/blog/introdu...
Introducing Nested Learning: A new ML paradigm for continual learning
We strive to create an environment conducive to many different types of research across many different time scales and levels of risk.
research.google
November 12, 2025 at 7:35 AM