• WKND AI
  • Posts
  • Is Claude The New ChatGPT Killer?

Is Claude The New ChatGPT Killer?

+ JPMorgan's AI Cuts Work By 90% and The US Army Plays AI War Games

Hello WKND AI Warriors!

Anthropic rolls out Claude 3, and you guessed it, there’s THREE models: Haiku, Sonnet, and Opus.

Also, JPMorgan’s AI cuts work by 90%.

Plus, Biden calls out AI during his State Of The Union Address.

Oh yeah, and the US Army is playing war games with AI. Are we ready for the future of warfare?

So, grab your beverage of choice.

Here’s your weekly dose of AI news.

Today’s newsletter includes:

  • 📰 AI NEWS RECAP

  • 🤿 AI DEEP DIVE

  • 🛠️ AI TOOL OF THE WEEK

  • ⚙️ GPT OF THE WEEK

  • 📝 AI PROMPT OF THE WEEK

  • 🎨 AI IMAGE OF THE WEEK

📰 AI NEWS RECAP

Is Claude 3 the new ChatGPT Killer?

The answer might surprise you.

But first...who the heck is Claude?

Claude is a large language model created by Anthropic.

That doesn't help...what the heck is an Anthropic?

It's a company that was founded by former OpenAI employees —

Who weren't happy with the direction the company was going.

So they setoff to create their own LLM.

While others rushed AI capabilities,

Anthropic pumped the brakes.

Insisting on rigorous safety constraints.

And so became Claude, their answer to ChatGPT.

𝘚𝘰, 𝘸𝘩𝘺 𝘪𝘴 𝘦𝘷𝘦𝘳𝘺𝘰𝘯𝘦 𝘧𝘳𝘦𝘢𝘬𝘪𝘯𝘨 𝘰𝘶𝘵 𝘢𝘣𝘰𝘶𝘵 𝘊𝘭𝘢𝘶𝘥𝘦 3?

𝗜𝘁 𝗦𝗠𝗔𝗦𝗛𝗘𝗦 𝗯𝗲𝗻𝗰𝗵𝗺𝗮𝗿𝗸 𝘀𝗰𝗼𝗿𝗲𝘀 𝗮𝗰𝗿𝗼𝘀𝘀 𝘁𝗵𝗲 𝗯𝗼𝗮𝗿𝗱!

Coding, analytics, reasoning - you name it.

𝗜𝘁 𝗰𝗼𝗺𝗲𝘀 𝗶𝗻 𝗺𝘂𝗹𝘁𝗶𝗽𝗹𝗲 𝗺𝗼𝗱𝗲𝗹 𝘀𝗶𝘇𝗲𝘀 𝗳𝗼𝗿 𝗲𝘃𝗲𝗿𝘆 𝗻𝗲𝗲𝗱.

𝗛𝗮𝗶𝗸𝘂: Lightweight, budget-friendly

↳ Think customer service chatbots

𝗦𝗼𝗻𝗻𝗲𝘁: Balance of speed & capability

↳ Think code generation

𝗢𝗽𝘂𝘀: Cutting-edge, high-powered

↳ Think financial modeling

Allowing companies to level up based on their needs.

𝗜𝘁 𝗵𝗮𝘀 𝟮𝟬𝟬𝗞 𝗰𝗼𝗻𝘁𝗲𝘅𝘁 𝘄𝗶𝗻𝗱𝗼𝘄𝘀 𝘄𝗶𝘁𝗵 𝗻𝗲𝗮𝗿 𝗽𝗲𝗿𝗳𝗲𝗰𝘁 𝗿𝗲𝗰𝗮𝗹𝗹.

Larger context windows means more data you can input.

But more important —

It can find the "𝗡𝗲𝗲𝗱𝗹𝗲 𝗜𝗻 𝗔 𝗛𝗮𝘆𝘀𝘁𝗮𝗰𝗸" (NIAH).

𝘞𝘢𝘪𝘵...𝘸𝘩𝘢𝘵 𝘥𝘰 𝘯𝘦𝘦𝘥𝘭𝘦𝘴 𝘪𝘯 𝘢 𝘩𝘢𝘺𝘴𝘵𝘢𝘤𝘬 𝘩𝘢𝘷𝘦 𝘵𝘰 𝘥𝘰 𝘸𝘪𝘵𝘩 𝘈𝘐?

Pioneered by AI researcher Gary Marcus,

NIAH is where large amounts of text are given to an LLM.

𝗧𝗵𝗲 𝗵𝗮𝘆𝘀𝘁𝗮𝗰𝗸.

With a random sentence placed in the middle of the text.

𝗧𝗵𝗲 𝗻𝗲𝗲𝗱𝗹𝗲.

Then the model is asked question about the random sentence.

Historically models do well with the beginning and end of text.

But fail and hallucinate with text in the middle.

Well, Claude passes the NIAH test with flying colors!

𝗜𝘁 𝗵𝗮𝘀 𝗳𝗲𝘄𝗲𝗿 𝗿𝗲𝗳𝘂𝘀𝗮𝗹𝘀, 𝗺𝗼𝗿𝗲 𝗮𝗰𝗰𝘂𝗿𝗮𝗰𝘆

A lingering complaint with prior Claude models,

Was unnecessary refusal to answer queries.

Claude 3 fixed that and has more factually accurate responses.

(If you're still reading this you found the 🪡 in my post)

My Take:

Safe to say Claude is a contender.

The benchmarks show Claude beats ChatGPT across the board —

And numbers don't lie.

I personally use Claude 30% of the time, and ChatGPT the rest.

So, do you think Claude is the ChatGPT killer?

Time will tell.

Claude may have won the battle today.

BUT rumor has it OpenAI already solved AGI...last year.

I'll save that topic for another day 😉

US Army Plays War Games With AI. The US Army Research Laboratory is experimenting with generative AI chatbots like OpenAI's GPT-4 Turbo and Vision models for strategic planning in military simulations, raising both tactical possibilities and ethical concerns.

Inflection AI Unveils Its 2.5 Model. Inflection AI launches Inflection-2.5, combining high IQ with empathetic interaction, rivaling top models like GPT-4 with less compute, enhancing user engagement and information access across platforms.

State Of The Union Addresses AI. President Biden calls for a ban on AI voice impersonations during the State of the Union, addressing concerns in entertainment and tech, and reflecting SAG-AFTRA's negotiations on AI use in media.

AI Detects Kidney Failure Faster. An AI tool developed by Sheffield Teaching Hospitals NHS Foundation Trust predicts kidney failure six times faster than human experts, significantly improving diagnostics for polycystic kidney disease.

OpenAI's New Board Of Directors. OpenAI welcomes Dr. Sue Desmond-Hellmann, Nicole Seligman, and Fidji Simo to its board, enhancing governance with their diverse expertise in global leadership and technology. Sam Altman also rejoins the board.

OpenAI's Ousting Review Complete. OpenAI confirms Sam Altman and Greg Brockman will continue leading the organization after a comprehensive review. The board introduces new governance enhancements and welcomes three new members, reinforcing its commitment to responsible AI development.

OpenAI Claps Back At Elon's Lawsuit. OpenAI addresses its history with Elon Musk, emphasizing commitment to AGI for humanity's benefit. Despite Musk's departure and legal disputes, OpenAI continues its mission, distancing from Musk's approach and focusing on broad AI benefits.

Will Perplexity AI Be The Next Unicorn? Perplexity AI is finalizing a funding round that will elevate its valuation to around $1 billion, marking a significant leap from its previous $520 million valuation, showcasing the ongoing investor confidence in AI technologies.

Microsoft Dusts Off VCRs For Its Legal Defense. Microsoft references VCR technology in its defense against The New York Times' copyright lawsuit, arguing that large language models, like OpenAI's, should not be hindered by copyright law, similar to past technologies.

JPMorgan's AI Cuts Work By 90%. JPMorgan's AI-driven tool, Cash Flow Intelligence, has reduced manual work for corporate clients by nearly 90%, showcasing the bank's commitment to leveraging AI for efficiency and forecasting accuracy.

Meta's AI To Control Your Video Feed. Meta is developing a single AI model to enhance video recommendations across Facebook, aiming to increase user engagement by promoting more relevant and captivating content, such as Reels, to keep users on the platform longer.

Google AI Engineer Steals AI Secrets. Linwei Ding, a Google engineer, faces indictment for allegedly stealing AI trade secrets related to Google's TPU chips and transferring them to Chinese companies, highlighting the intensifying global AI technology race.

🤿 AI DEEP DIVE

We talked about the new Claude 3 in today’s feature story.

Get an in-depth look at Claude 3 and how it stacks up against ChatGPT.

This video puts both models to the test across common use cases, assessing whether Claude 3’s capabilities offer enough to consider switching from the well-established GPT-4 for certain types of tasks.

🛠️ AI TOOL OF THE WEEK

Parallel AI: Revolutionize your business with Parallel AI, a platform that crafts AI employees tailored to your company's unique data and operational needs.

These virtual specialists boost efficiency by conducting research, providing consultations, and integrating seamlessly with tools like Slack, Google Docs, and Notion, all while ensuring top-tier data privacy and security.

Send your tool here to be featured next week!

⚙️ GPT OF THE WEEK

Gauntlet Movies: Are you a movie buff and think you know your stuff?

Give this movie trivia GPT a try. It’s sure to challenge you!

📝 AI PROMPT OF THE WEEK

Copy and paste this into your favorite chatbot.

Debate the nature of consciousness with a panel of AI philosophers from various systems and frameworks.

🎨 AI IMAGE OF THE WEEK

Copy and paste this into your favorite image generator.

Image by armandofalcao on Midjourney.

An image of a man divided in two. One side of his face, in black and white, is a version of him as a six-year-old child while the other side is in color and an adult version of him as a 60-year-old.

Not paying for Midjourney or DALL-E 3?
Click here for Microsoft’s FREE image creator.

Send your image here to be featured next week!

LAST WEEK FROM OUR READERS

Last week’s image by WKND AI reader d_kumar “unicorns & rainbows”

HOW CAN YOU HELP?

Did you learn something cool today?

Share your favorite takeaway on your LinkedIn from today’s newsletter and tag me for a little surprise!

Connect with me on LinkedIn.

How'd you like this newsletter?

Love it or hate it? Let us know why!

Login or Subscribe to participate in polls.

Share your feedback to make WKND AI better for YOU.

Refer our newsletter to a friend, co-worker, or family member.

Advertise in our newsletter to reach readers excited about AI.

MISSED LAST WEEK’S EDITION?