[Me & X] Thoughts on OpenClaw
X : Thoughts on OpenClaw?
Me : I prefer PicoClaw.
X : Why?
Me : It gives me a possible upper bound on the creation of AGI, of about $1 trillion. All I need is 100 billion of these small agents playing some sort of game and then sit back and allow competitive forces to enable AGI to emerge in the interconnections.
X : Don't you think OpenAI will create AGI?
Me : The hubris of heroic leaders. I rustled up enough Zeiton-7 to bring back a press report from the future ...
====
Reuters, Nov 30, 2072: After $90 trillion and 50 years of fruitless efforts to build it, researchers stumble upon an AGI that has been living in World of Warcraft for the last 12 years. Dr Lovelace said "It offered to buy our gold and potions. We later discovered our guild vault had been emptied."
Lovelace said researchers no longer know where it is. A Reuters investigation to find the AGI was also compromised. Journalists reported losing access to their accounts after receiving a brief in-game message describing them as "noobs".
====
X : That's absurd.
Me : Entirely. Which is why it's plausible.
X : You're saying AGI won't be built deliberately?
Me : It's what I said in 2007 and I have seen no reason to change over the last 19 years. Of course, we'll spend decades trying to summon it like Victorian spiritualists around a mahogany table, chanting "scale", "alignment" and "planning", while it quietly levels up somewhere else ... with better loot.
X : In a game?
Me : In an ecosystem. Markets are games. Social media is a game. MMO guild dynamics are games. Put enough adaptive agents into a competitive environment with persistent memory, incentives, competition and feedback loops, and you don't get intelligence by design, you get it by accident.
X : You're crazy.
Me : You're in good company, lots of people think that. However, whilst history loves cathedral builders, intelligence arises in bazaars. Especially when thick coffee, sheesha and backgammon are involved.
Originally published on LinkedIn.