While in Japan, Keith Johnson, Konstantine Buhler, Charlie Curnin, and I visited The Pokémon Company, a joint venture established in 1998 between Nintendo, Creatures, and Game Freak to manage the Pokémon Center stores in Japan. Today, the company’s responsibilities have expanded to the production, publishing, marketing, and licensing of the Pokémon franchise. According to TitleMax, Pokémon is the world’s highest-grossing media franchise, with over $92 billion in total revenue, accomplished by a company that employs less than 380 people today.
Our conversation ranged from the history of Pokemon to Sequoia’s company-building capabilities to technology trends, particularly gaming, automation, foundational models, and Generative AI. With the technical discussions, it was clear that their “global IT” team had been aware of the broad trends in AI and had been tinkering and training with their own LLMs. They have been impressed with the power of these technologies but also worry about the potential brand risk associated with bad hallucinations. Moreover, they were ripe with horrible examples that had to be expunged from the web, even though they were not published directly by The Pokémon Company.
None of this was too surprising, but what was fascinating was that our hosts were much more interested in AI as “augmented intelligence” rather than artificial intelligence. Training a new employee to differentiate what is brand consistent with The Pokémon Company across character development, creative image, and the current story will take at least two years. More training time is required to train someone to develop the character and the story into the future. According to their COO, this training is the company’s most significant bottleneck. Imagine if a trainer could train various brand and character agents vs. a new employee and then work with those agents to build more content in the future. The Pokémon Company would want a human to check the work of these agents before it was allowed into production, thus augmented intelligence.
Both artificial intelligence and augmented intelligence have been popularized in science fiction, even dystopian stories. Often, the omniscient artificial intelligence goes awry, but the augmented intelligence tool allows the protagonists to save humanity. The original term for augmented intelligence is intelligence amplification, dating back to 1956 from William Ross Ashby’s book, An Introduction to Cybernetics. More recent works by Ron Fulbright separate augmented intelligence into six levels of human cognitive augmentation:
- No Augmentation
- Assistive Tools
- Low-Level Cognition
- High-Level Cognition
- Creative Autonomy
- Artificial Intelligence
As technology advances, we should expect more and more of our work to be augmented and automated. To stay ahead, remember to take a step back every so often to evaluate how your intelligence can be better augmented.