A positive future of work, abundant expertise, digital twin elicitation, and more

📖In this issue

• A prosperous future of jobs: Redux

• An “agent bank” of 1000 accurate replicas of real humans

• What “abundant expertise” means for strategy

• Coaching is a major enterprise GenAI use case

• Kevin Clark & Kyle Shannon on collective intelligence, digital twin elicitation, data collaboratives, and the evolution of content

• The model leapfrogging and improvement continues

💡A prosperous future of jobs: Redux

In May I shared my Future Job Prosperity report. Given the current very negative outlook on the future of jobs I thought it was worth going back to it. At this point I don’t think it needs major revisions, I just made a few changes.

Please have a look and let me know how I can bolster and improve the case, including with cogent, specific arguments against these 13 points.

1. The potential of Humans + AI

2. AI enhances value-generating skills

3. Creation of new jobs

4. Unique human capabilities at the fore

5. Specialization reduces substitutability

6. Enhanced education and learning

7. Comparative advantage

8. Attraction of talent

9. Work redesign

10. Humans’ extraordinary adaptability

11. Emotional intelligence and human connection

12. Preference for human work

13. Design for inclusive prosperity

🧠🤖Humans + AI

An “agent bank” of 1000 accurate replicas of real humans

Researchers have generated an "agent bank" of over 1000 AI agents that each accurately simulate a real human. Extended interviews and effective agent design enabled 85% predictive accuracy for replicating attitudes and behaviors using the General Social Survey.

The agents replicated humans results in most behavioral experiments, with effect sizes showing a correlation of 0.98 to human participants, who themselves showed a 0.99 internal consistency.

Paper: Generative Agent Simulations of 1,000 People 
Discussion and implications

What “abundant expertise” means for strategy

A very interesting article in Harvard Business Review looks at strategy in a world where "the overall body of expertise in the world is constantly expanding... [and].. the cost of accessing expertise is constantly falling."

I have a slightly different take from their analysis, but framing the shift in strategy in this way is extremely useful. This has to be a central question for every leader and strategy executive.

Strategy in the Era of Abundant Expertise
Summary, discussion, and reframed questions

Coaching is a major enterprise GenAI use case

A very interesting report from VC Menlo Ventures on the State of Generative AI in the Enterprise has plenty of interesting data. Among other titbits, spending on GenAI has increased over 6x over the last year, and RAG is rising rapidly as the primary architectural approach, with fine-tuning and prompt engineering used far less.

One thing that struck me is that coaching ranks 8th on the enterprise use case list, behind very obvious ones like coding, support chatbots, and meeting summaries. This is in fact a very powerful human augmentation application. But I’m surprised and greatly encouraged that companies are recognizing this.

2024: The State of Generative AI in the Enterprise
Summary and discussion

🎙️This week’s podcast episode

Why you should listen

In this conversation Kevin Clark and Kyle Shannon share in detail how they have built an array of digital twins of insightful thinkers, and how they use them to think better and generate useful content. It’s a pragmatic and fun conversations.

💡Reflections

The model leapfrogging and improvement continues

All the talk these days is of LLMs hitting a wall in progress. That looks likely to be true if we compare wiht the breathtaking baseline pace of the two years up until a couple of years ago. But it doesn’t mean that there isn’t progress.

Google latest Gemini experimental model has leaped ahead on just about every measure, beating o1-preview, 4o, and 3.5 Sonnet as well as a crop of open source models that are proving to be highly competivie with the commercial models.

No doubt OpenAI and likely Anthropic will again take the lead before long, every change in leader representing an improvement in capabilities.

Innovation hasn’t stopped, and massively funded intense competition is driving advances.

We may need to wait for a major algorithmic advance to match the pace of the last two years, but for now things are still moving ahead at a solid pace.