Top ideation prompts, one-person billion dollar companies, Mobius leadership, soaring LLM price-performance, and more

April 06, 2024

“AI is going to change everything, but it’s going to be a partnership between humans and AI, not a competition.” - Mark Cuban

We are one quarter the way through 2024 and generative AI capabilities are already substantially ahead of where their were at New Year.

While most are focusing on LLM progress and how it is being deployed in organizations, fortunately the attention to ‘Humans + AI’ thinking is increasing. Leaders and individuals are realizing we need to work on the role of AI relative to humans, to enhance not replace them.

Ideation is one of the most powerful applications of generative AI. We share below a number of very useful ideation prompts and approaches.

Ross

 

📖In this issue

  • Top ideation prompts

  • Emotionally responsive AI voice conversation

  • Is a one-person billion-dollar company possible?

  • Apple is reported to be exploring home robotics as its ‘next big thing’ after closing its car project

  • Excellent visual explanation of Transformers

  • Charles Hampden-Turner on Mobius leadership, reconciling paradoxes, dilemma strategies, and conscious capitalism

  • LLM price-performance is rising by orders of magnitude

 

 

🖼️ Framework

Ideation prompts

Among the many very interesting discussions on the Humans + AI Explorers Community is a thread on ideation prompts. It includes:

  • A prompt for generating 100 ideas, improving them, and selecting the best, adapted from the “Prompting Diverse Ideas” paper by Meincke et al.

  • Starting by asking for bad ideas

  • Building inferred personas of everyone in a workshop and using them for ideation processes

  • Using these to generate complementary personas for creative abrasion

  • Sequences of prompts for hypothesis-led trend scanning using Jungian archetypes

 

👩🤖Humans + AI update

Emotionally responsive AI voice conversation

Hume AI is getting a lot of attention for its “AI with emotional intelligence”, which interprets emotion in voices responds empathically. Others seem to have had a better experience than I did, it is not yet very fluid but shows the promise of soon being a highly empathic interlocutor. Try the demo.

Is a one-person billion-dollar company possible?

This article examines whether this is realistically possible rather than hype, with some interesting analysis and perspectives. It concludes that it is “incredibly challenging, if not impossible”, but that very profitable companies could be run by one person assisted by AI.

Billie Eilish, Katy Perry and others object “irresponsible” AI

“When used irresponsibly, AI poses enormous threats to our ability to protect our privacy, our identities, our music and our livelihoods… Some of the biggest and most powerful companies are, without permission, using our work to train AI models. … For many working musicians, artists and songwriters who are just trying to make ends meet, this would be catastrophic.”

 

🔥Hot news in AI

  • Apple claims its new ReaLM model that runs on-device “substantially outperforms” GPT-4 - ZDNet

  • Apple is also reported to be exploring home robotics as its ‘next big thing’ after closing its car project - CNBC

  • ChatGPT no longer requires an account - TechCrunch - and paid users can edit DALL-E images - Axios

  • The visual explanation channel 3Brown1Blue has created an excellent explainer of LLMs which is attracting some buzz, worth watching below:

 

🎙️Latest podcast episode

Charles Hampden-Turner on Mobius leadership, reconciling paradoxes, dilemma strategies, and conscious capitalism

What you will learn

  • Exploring the genesis of “Maps of the Mind”

  • The power of paradox in understanding the human mind

  • Reflecting on a career; tying together themes of management and leadership

  • The Mobius strip as a metaphor for solving complex problems

  • Addressing societal polarizations through integrated thinking

  • The role of conscious capitalism in today’s business world

  • Visualizing paradoxes; the use of imagery in comprehending complex ideas

 

💡Resources and insights

LLM price-performance is rising by orders of magnitude

The price-performance of LLMs is orders of magnitude better than just a couple of years ago, driven by innovative architectures and more efficient training and inference. So what's next?

This excellent analysis by Anthropic's Karina Nguyen plots log scale cost on the y-axis against performance on the Massive Multitask Language Understanding (MMLU) benchmark.

The MMLU is designed to be similar to tests given to humans, which includes a variety of tasks including knowledge and reasoning across domains. Some tests have suggested non-specialist humans would achieve 35% while specialists would hit 90% in their area of expertise.

Chare