Humans + AI interfaces, the human superorganism, better prompting strategies, and more

One of the most important ways we will tap the potential of Humans + AI is through better interfaces. The magic of LLMs is that we can interact directly with AI. Yet more sophisticated prompting techniques can generate far better results.

As an Amplifying Cognition newsletter subscriber you are first to get access to our new app ThoughtWeaver, initially available only in our BYO Key plan. You can get a 30 day free trial using the code ACNEWSLETTER.

It provides a highly efficient interface to access our deep library of crafted assistants, prompts, and contexts, or you can use your own preferred prompts. I’d love to hear what you think and how we can make it more useful for you.

Ross.

• Strategies for better prompting results

• Hallucination is a feature not a bug

• AI makes new mathematics discovery

• Byron Reese on the human superorganism and collective intelligence

• Attention and awareness for organizations

Six strategies for better prompting results

OpenAI has published a simple guide to prompt engineering with six basic strategies. It is similar to other guides it has published before but provides a current and distilled approach, including detailed examples. Useful for users of LLMs at all levels.

Hallucination is a feature not a bug

Top AI researcher Andrei Karpathy notes that “in some sense, hallucination is all LLMs do. They are dream machines. We direct their dreams with prompts… It's only when the dreams go into deemed factually incorrect territory that we label it a "hallucination". It looks like a bug, but it's just the LLM doing what it always does.”

On the "hallucination problem"

Axel Springer partners with OpenAI on journalism

German media giant Axel Springer has made a deal with OpenAI to give them access to selected news, training their models with real-time reputable reporting, and using the models to enhance their journalism. There are few details at this point, the interesting part being the specific value to Axel Springer’s news reporting beyond the financial aspects of the deal.

Partnership with Axel Springer to deepen beneficial use of AI in journalism

GoogleDeepMind makes discovery in mathematics theory

A new AI method for searching new solutions in mathematics and computer science, FunSearch, has made a novel mathematics discovery, the first time an LLM has done so. While this was a domain-specific case, this suggests that the frontiers of math could be extended at an accelerating rate using AI tools.

FunSearch: Making new discoveries in mathematical sciences using Large Language Models

Attention and awareness for organizations

A piece in the excellent Flux Review newsletter provides a useful framework for thinking about attention vs awareness.

Thank you for reading!

Please do let us know any feedback and what you’d like to see in the newsletter. We’ll be continuing to develop the content to be better and more useful for you in coming weeks and months.