Shelly Palmer - Opening AI's Black Box

Greetings from Terminal 4 at JFK. I'm heading up to Syracuse this morning to teach my first class of the semester in person. As I'm waiting for this flight, I've been reading the resumes of my master's students – what an awesome group!

In the news: Anthropic (the company behind ChatGPT competitor Claude) has taken an interesting step toward AI transparency by publishing Claude's system prompts – the ones that shape the model's behavior. These prompts dictate not only the model's restrictions (like avoiding facial recognition) but also their intended personality traits, such as intellectual curiosity and impartiality. According to Anthropic, its goal is to position itself as an ethical leader in AI. That would certainly set it apart from the pack.

It's nice to see a big foundational model builder make an effort to demonstrate the importance of human oversight. It's also nice to understand what kinds of prompts are necessary to get a foundational model to behave as expected (which require system prompts, fine tuning, and a bunch of other post-training techniques).

Take a minute and read the whole prompt by clicking the "July 12, 2024" box on this page. It's fascinating reading.

As always your thoughts and comments are both welcome and encouraged. Just reply to this email. -s

P.S. Thanks for all the info and tips about my upcoming trip to Bangkok. I'm excited for next week's trip. If you've got any more "must see" or "must do" recommendations, please LMK.

Today's Most Interesting Stories

 

Shelly's Blog: Making Your LLM Yours: Enhancing LLMs with External Signals

Big foundational models like GPT-4, Gemini, Claude, and Llama are awesome, but they are not experts in your business. One of the most effective ways to augment an LLM is with the addition of external signals, which are various types of real-world data and information that can be used to influence and improve the performance and relevance of LLMs. Let's take a high-level look at how external signals can transform LLMs, making them more relevant, responsive, and ultimately, more useful.

Read More · Shelly Palmer

OpenAI, Adobe and Microsoft support California bill requiring watermarks on AI content

OpenAI, Adobe and Microsoft have thrown their support behind a California bill requiring tech companies to label AI-generated content, according to letters from the companies viewed by TechCrunch. The bill is headed for a final vote in August.

Read More · TechCrunch

Samsung will soon let you create AI wallpapers on its touchscreen refrigerators

Samsung has announced new AI capabilities for its smart appliances, including new AI-generated wallpapers for certain refrigerators, as well as an AI upgrade for its Bixby voice assistant.

Read More · The Verge

 

Apple is expected to debut the first generative AI iPhone at its September 9 event

Apple has announced the date of its next major event, where the iPhone 16 is expected to launch.

Read More · CNN

Nvidia partner says it can cut data center energy use by 50% as AI boom strains power grid

Data center company Sustainable Metal Cloud wants to reduce the strain on energy resources and carbon emissions.

Read More · CNBC

Amazon aims to launch delayed AI Alexa subscription in October

The revamp of the personal voice assistant, which includes new AI-enhanced features for cooking, kids, and news, comes just weeks before the US presidential election.

Read More · Washington Post

Today's Most Interesting Stories

 

Shelly's Blog: Making Your LLM Yours: Enhancing LLMs with External Signals

Big foundational models like GPT-4, Gemini, Claude, and Llama are awesome, but they are not experts in your business. One of the most effective ways to augment an LLM is with the addition of external signals, which are various types of real-world data and information that can be used to influence and improve the performance and relevance of LLMs. Let's take a high-level look at how external signals can transform LLMs, making them more relevant, responsive, and ultimately, more useful.

Read More · Shelly Palmer

Apple reportedly announcing iPhone 16 lineup and more on Sept. 10

Apple will be unveiling new products on September 10, with the announced phones going on sale on September 20, according to a report from Bloomberg’s Mark Gurman.

Read More · TechCrunch

Meta Cancels High-End Mixed Reality Headset After Apple Vision Pro Struggles

Meta has canceled work on a high-end mixed reality headset that it was developing to compete with the Apple Vision Pro, reports The Information.

Read More · MacRumors

‘Megalopolis’ Cuts Ties With Marketing Consultant Behind Trailer Debacle; Fabricated Critic Quotes Were Generated By AI

'Megalopolis' has parted ways with marketing consultant Eddie Egan following the recent trailer debacle, Deadline has confirmed.

Read More · Deadline

AI Could Make Coders Obsolete in Two Years, AWS Chief Predicts

Amazon Web Services CEO Matt Garman is not the first to envision AI doing the bulk of software development.

Read More · Decrypt

Google Wallet now works for California driver’s licenses

Android users participating in a pilot program can now add their California driver’s licenses to Google Wallet. The “mobile driver’s license” program is currently open to 1.5 million Californians, and previously only allowed digital IDs in the state’s “CA DMV Wallet” app for iOS and Android.

Read More · The Verge

 

ABOUT SHELLY PALMER

Shelly Palmer is the Professor of Advanced Media in Residence at Syracuse University’s S.I. Newhouse School of Public Communications and CEO of The Palmer Group, a consulting practice that helps Fortune 500 companies with technology, media and marketing. Named LinkedIn’s “Top Voice in Technology,” he covers tech and business for Good Day New York, is a regular commentator on CNN and writes a popular daily business blog. He's a bestselling author, and the creator of the popular, free online course, Generative AI for Execs. Follow @shellypalmer or visit shellypalmer.com.