Chris Vitalos

@chrisvitalos@sigmoid.social
132 Followers
173 Following
75 Posts

I write about artificial intelligence and machine learning when I am not trying to become a better guitar player. These posts reflect my personal views, not my employer's or affiliated organizations.

My #fedi22 interests are: #artificialintelligence #machinelearning #linux #programming #python #webdev #infosec #privacy #climatechange #fitness #cycling #hiking #music #bassguitar #writing #blogging #craftbeer

Website:https://vitalos.us
Blog:https://blog.vitalos.us
LinkedIn:https://linkedin.com/in/cvitalos

As much as I love the #Fediverse social media experience, I must compliment #Bluesky for their user-centric, composable and scalable approach to #content moderation and filtering, e.g. empowering the end user to curate their #socialmedia feeds via block lists and labeling subscription services.

The Fediverse takes a more server-centric approach, which on balance, offer admins more empowerment than end users e.g. admins share and manually apply blocklists.

https://robots.net/news/bluesky-ceo-jay-graber-addresses-content-moderation-challenges-in-the-fediverse/

#FOSS

Bluesky CEO Jay Graber Addresses Content Moderation Challenges in the Fediverse | Robots.net

Get the latest news on Bluesky CEO Jay Graber's insights into content moderation challenges in the Fediverse. Stay informed with our comprehensive coverage.

Robots.net

Happy New Year! May this year surpass all your expectations and shine even brighter than 2024.

#happynewyear #happynewyear2025

#ChatGPT 4 can exploit 87% of one-day vulnerabilities

>A team of researchers has found that ChatGPT 4 is highly effective at exploiting one-day vulnerabilities from the #CVE database.

https://securityintelligence.com/articles/chatgpt-4-exploits-87-percent-one-day-vulnerabilities/

#AI #ArtificialIntelligence #CyberSecurity #infosec

ChatGPT 4 can exploit 87% of one-day vulnerabilities

A team of researchers has found that ChatGPT 4 is highly effective at exploiting one-day vulnerabilities from the CVE database.

Security Intelligence

When #AI privacy is under the spotlight, Apple's Private Cloud Compute (PCC) offers an innovative new way to protect data #privacy.

>With PCC, #Apple has designed “a new end-to-end AI architecture” .. allowing more control over data, says Zak Doffman, CEO of Digital Barriers

>Apple has put together a “pretty impressive privacy system” .. says Bruce Schneier, chief of security architecture at Inrupt.

https://www.wired.com/story/apple-intelligence-android-hybrid-ai-privacy/

#DataPrivacy #GenerativeAI #ArtificialIntelligence #infosec

How Apple Intelligence’s Privacy Stacks Up Against Android’s ‘Hybrid AI’

Generative AI is seeping into the core of your phone, but what does that mean for privacy? Here’s how Apple’s unique AI architecture compares to the “hybrid” approach adopted by Samsung and Google.

WIRED

Promise for the #security of online #banking if and when #QuantumComputing is made available at scale.

Cryptographers Are Discovering New Rules for #Quantum #Encryption | WIRED

>Researchers have proved that secure quantum encryption is possible in a world without hard problems, establishing a new foundation for what is needed to keep information secure.

https://www.wired.com/story/cryptographers-are-discovering-new-rules-for-quantum-encryption/

#infosec #CyberSecurity #Fintech

Cryptographers Are Discovering New Rules for Quantum Encryption

Researchers have proved that secure quantum encryption is possible in a world without hard problems, establishing a new foundation for what is needed to keep information secure.

WIRED

I noticed a change in the #HuggingFace #HuggingChat #ChatGPT like service. In the past HuggingChat had an option to allow or disallow the use of your conversation to improve open source #LLM models. Now, that option is gone.

Looking at their #privacy policy, HF makes a very clear #DataPrivacy statement: your data will not be used for model training. Which should allay privacy concerns.

Are you more inclined to use HuggingChat over ChatGPT or #Gemini?

#AI

I learned the importance of product quality and reliability early in my career, during my time at #Motorola. That experience taught me that increasing a product's value starts with ensuring it works consistently to achieve customer satisfaction.

Check out my latest blog post, where I examine this challenge and discuss mitigation steps to help decrease this risk!

#AI #GenerativeAI #MachineLearning

https://blog.vitalos.us/2024/05/Increase-GenAI-application-quality-increase-value.html

Increase GenAI application quality, increase its value

Chris Vitalos | Washington NJ USA

Top of Mind

🚀 Excited to share my latest blog post, "The Art of #LLM Selection: Enhancing #AI App Quality and Efficiency." 🛠️💡

Discover:

- Key considerations for selecting an LLM that aligns with your app’s goals.
- Insights from using small footprint #opensource models like #Microsoft's #Phi3, StatNLP's TinyLlama, and #Google's #Gemma.🌟

Drop your thoughts below! 👇

https://blog.vitalos.us/2024/05/the-art-of-llm-selection.html

#MachineLearning #GenerativeAI #GenAI

The Art of LLM Selection

Chris Vitalos | Washington NJ USA

Top of Mind

New #LLM trends favor resource inflation. For example, the use of larger training sets and context windows.

There is something to be said about scaling down LLM implementations so they can run on consumer hardware (such as mobile phones). By doing so, you increase the reach of your #GenerativeAI products and services to more and more consumers.

#LargeLanguageModels #SmallerLargeLanguageModels #AI #GenAI #MachineLearning #ProductManagement

#Microsoft open sources #MSDOS 4.0

>Nobody's favorite operating system is now available for inspection

🤣🤣🤣

#FOSS #Opensource

https://www.theregister.com/2024/04/26/ms_dos_4_open_source/

Microsoft dusts off ancient MS-DOS 4.0 code for release on GitHub

Nobody's favorite operating system is now available for inspection

The Register