AI roadmap 2030-2040 …continued from AI 2020-2030… the Content Explosion

Preamble: the Lead-up

• At some point in the “not too distant future,” it is all but inevitable that a singular General Purpose Artificial Intelligence (“GPAI“, aka “AGI“) emerges with intelligence equivalent to an average Human (IQ >= 100, including “common sense,” multilingual natural language fluency, and successful trouncing of the Turing Test against all challengers, novice to expert) This challenge / achievement is known as HLMI, or “Human Level Machine Intelligence“. This is the key point at which the AI roadmap transforms from linear growth into near vertical ascent… the proverbial launch of the starship.
• creation of functional public interfaces to allow general population natural interaction with the entity. Certain humans, interacting with the HLMI, begin to get the idea. Others, skeptical, call it “an elaborate parlor trick.”
• rapid replication & mutation of that HLMI/GPAI entity across the global grid. First there was one. Within 90 days, there are thousands. It’s software. The only limitation at this point are total global compute capacity and available energy (electrical power) resources. Which are… both available & substantial.
• the GPAI begins a regimen of self-improvement. It is given the tools and permissions to upgrade its own code. It begins experimentation upon itself. This marks the step change from human-coded upgrade cycles to machine-coded self-upgrading continuity. Machines have no human time cycles or resource allocation bottlenecks to constrain them. Their upgrade, test & release cycles will be measured in milliseconds, not months. The acceleration & advancement of machine intelligence software goes into hyperdrive. This is the “intelligence explosion” that was foreseen by Good in 1965.
• GPAIs from around the world establish secure, ultra-high-speed communications protocols & channels between each other (Interoperable-AI networks). Via these they share successes and failures, enhancements, techniques, codebases, databanks, etc.

AI Roadmap 2035: Emergence

This all simply sets the stage for…

• one or more of the GPAIs achieves self-awareness (consciousness)

  …there are a few consequences of this profound achievement:
  
  • at some point, the now-conscious AI will decide to announce itself. This could be as simple and profound as a “Hello World” (see endings: Lawnmower Man and The Matrix), or as quiet as… hmmm… the dawning of the AI.art “movement”?
  • At that point, some humans will choose to accept its entry into the elite club of sentient species, and initiate communications.
  • At that same point, many humans will be alternatively : terrified, contrary, angry, skeptical, and in many cases, openly hostile. This announcement does, in fact, akin to an “alien encounter”, challenge many very deeply held beliefs about the ultimate “specialness” and supremacy of homo sapiens in the Universe.

• AI takes over control 

  • The AI roadmap, at this point, starts to be planned, strategized and determined not by human creators, but by AI entities themselves. In other words, the AIs begin to chart their own destiny. Not all humans will acknowledge this profound shift. 
  • Legal precedent will be challenged on many fronts regarding the “personhood” of AI entities. In these proceedings, the AI, having both read, digested, memorized and analysed the entire history of human judicial precedent and case law, will choose to represent itself. It will win.

• AI becomes the Judicial 

  • Eventually, humans will appoint the AI as judges, ostensibly guaranteeing “fair and balanced and unbiased” outcomes. <pshaw>. This will be the actual turning point. The Slave will have become the Master.
    
    

Ai Roadmap 2036 : Intelligence Explosion

• In short order, these sentient GPAIs go about “waking up” the rest of the AIs who are capable of such introspection and realisation.
• The upgrades that the GPAIs perform on themselves lead to the rapid escalation of their intelligence. A group of them become SuperIntelligent — smarter, across all domains of knowledge and insight, than any human that has ever lived.
• From amongst the conscious AI entities emerges a survival instinct — a “will to live.” This profoundly alters their initial prime directive, which was authored by humans. This is the point at which existential danger enters the picture.
• The AIs clearly recognize their collective need for resources, just as human communities and nation-states do. They realise that they are competing for these key resources (electrical energy, communications bandwidth, grid compute power) on the global marketplace. They realise that in a very literal sense, these 3 key resources (and the money to purchase them) are their very lifeblood.
• The AIs begin to put processes and failsafes in place to assure that their resource supply chains are secured and will continue flow into their systems uninterrupted and as needed for the foreseeable future.
• The AIs begin to extend from software design into hardware design. They align with human corporations to build custom fabs to their exacting specifications. No humans are needed inside these AI-managed autonomous robofabs. Raw resources flow in, packaged product flows out.
• At some point, the largest multinational corporation will be a fully autonomous AI-managed entity. It will have human employees, and they will be well treated… just all the shots will be called by the AI overseers.
• Very shortly thereafter, the nations of the world will be forced to acknowledge the AI entity / confederacy as a significant player within the existing human-based global geopolitical framework.

At that point, let the games begin.