📈 How are companies using AI?

➜ “Unbelievable potential. You’ve gotta get it, make it part of the management conversation” – Jamie Dimon, one of the most influential figures in finance, once again endorsed the potential of AI. Dimon is a longtime CEO at JPMorgan Chase & Co. where AI is being used across functions like risk, fraud, marketing, and customer relations.

Bloomberg

➜ According to Gartner,  41% of chief audit executives are using, or plan to use, generative AI in 2024. To be honest, not sure how they managed without it! If you’ve ever written (or read) audit reports, you know exactly what I mean.

Digit News

➜ Generative AI in insurance? FWD Group, a Hong Kong-based insurer, just came together with Microsoft to squeeze as much as possible out of Azure OpenAI and AI models. The time for GenAI in customer service, marketing, underwriting, and claims operations is NOW.

Computer Weekly

➜ Go big or don’t go at all: Wolters Kluwer Health will soon be synonymous with clinical decision support AI. Their AI Labs, a GenAI beta launched in 2023 and powered by UpToDate Enterprise™ Edition, now offers over 100 U.S. hospitals and health systems access to AI Labs’ testing and validation program. I’m talking about an LLM trained on continually updated, trusted content. Any doc will tell you – that is massive. 

Yahoo Finance

➜ Humanoid robots are definitely having a moment: a Bay Area-based startup has just confirmed a $675 million Series B round. 

TechCrunch

➜ A prompt is all you need to ship 3D animated websites without coding. That’s the promise behind Dora. The company opened whitelisted access to the tool. When I signed up last Friday, I was #282,368 in line. 

Dora.run

💡 What are the nerds up to?

➜ You can now train a 70 billion parameter language model at home on a regular desktop computer with gaming GPUs. Answer.AI, a for-profit R&D lab, has pulled it off by combining two techniques – QLoRA (Quantized Low-Rank Adaptation) and FSDP (Fully Sharded Data Parallel). As Dana Majid, CTO at Bird put it “Something big happened in open source AI land past week.”

Answer.AI

➜ Turns out LLMs may be aware of being tested and evaluated.“I suspect this pizza topping "fact" may have been inserted as a joke or to test if I was paying attention” – noted Opus, an LLM from the Claude 3's new family, in reaction to researchers' prompts. Full story on X.

X

➜ “AI Prompt Engineering Is Dead” claims Dina Genkina. Obvious clickbait, worth the read though. “Don’t bother, there is no trend” say researchers to all these prompt-engineering guides, cheat sheets, and advice threads on social media aimed to help you get the most out of ChatGPT.

Spectrum IEEE

➜ JetBrains unbundled their AI Assistant, which is now available as a separate plugin. Now, it is up to a developer if, and when, to use AI-powered technologies.

JetBrains

➜ Faster and more efficient AI models are on their way. Two groups of computer scientists claim to have found a new way to multiply large matrices faster than ever before. Matrix multiplication is a key component of AI models like ChatGPT.

ArsTechnica

➜ “I see the danger of this concentration of power to proprietary AI systems.” If you can spare 3 hours, I recommend tuning into this conversation with Yann LeCun, Chief AI Scientist at Meta. As a bonus, an interesting comparison made by LeCun on the importance of learning AI from high-bandwidth sensory inputs such as vision: “In a mere 4 years, a child has seen 50 times more data than the biggest LLMs trained on all the text publicly available on the internet.”

YouTube

➜ Dave Wilding built a cool (and simple!) AI writing assistant to help him improve his Chinese. Seems that pairing GPT-3.5 with Microsoft machine translation works miracles. 

Unscrambler

➜ To finish up: Seven teachers across the world sharing their insights on AI tools for educators. Longer read, but worth every minute.

The MarkUp

Thanks for reading!

I share these stories because I believe it's important for all of us to keep up with AI. To support my mission, share this newsletter on LinkedIn.