Interlude 2

In 2051, the first true AI had successfully been initiated, after 70 years of failure. It was an accident, as many new technologies are.

Limited neural nets and genetic algorithms working in concert had been used for decades to permit more responsive games, but as unlimited standalone beings they had always failed, to the dismay, fury, and puzzlement of AI researchers. AI's with the comparative intelligence of a dog were possible and stable. Anything more intelligent was not.

Dr. Alma Gordon, of UCLA, one of the leading psychologists of the period.  Dr. Jamil Palanirajan, Bhagwant Institute of Technology, one of the acknowleged leaders in AI research. Both in their early 30's. Both tired after presenting controversial papers to their respective conferences. Both in the hotel bar, having a drink to wind down. Both going outside for a highly illegal tobacco cigarette at the same time. Dr. Palanirajan's lighter fails. He asks Dr. Gordon for a light. Light discussion. Back to the bar, share a table and continue to talk. Dinner. Mutual attraction. Bed. Courtship. Partnership. Of such chance meetings the future is made.

15 years of solid work, tragedies and triumphs interlocking. Dr. Gordon's insight that a human type intelligence will dissociate once it reallises that it can only ever take orders was the key. The more intelligent the organism, the more likely that it will ask "what's in it for me."  So remove the concept of self. The AI they finally built was intelligent, about genius level in human terms, but decidedly not human in it's responses. 

Dr. Gordon and Dr. Palanirajan moved their attention to the problem of human thought and memory. Using the various false starts and leads from AI research, within the year they and their research team had produced the outline of what would become the ghosting process.

AI's were quickly taken up by industry - they were after all fast, efficient, and did not need rigorous programming to provide answers, as they had intuition of a sort. Within 6 years every government, business, spaceship and research facility had it's own AI. Mankind had an ally to help clean up the mess previous generations had made. The whole problem too complex to be analysed other than by an AI.

And then things began to go wrong.

Small at first - a few deaths here and there as a result of AI problem analysis. But deaths increasing in frequency. The odd brushfire war, which on quiet investigation was found to be caused by AI recommendations. The problem, of course, was the maximum survival of life.

AI's had, as part of their hard coding, an innate respect for life. Not human life, just life. They had respect for humans also hard coded - but at a lower priority. One of the unanticipated failures of Asimov's Laws, as interpreted by a truely intelligent machine.

Humanity rebelled. AI's were purged all over the planet. An intelligence that did not put man as the pinnacle of life was simply not to be permitted. Off planet, where they could do no harm - fine. The mob could not touch them, and they could not touch the mob.

The ghosting process by then had been used for 12 years. Over 50,000 people had been ghosted. Several ghosts were also terminated, when a mob gets together the fine shadings of right and wrong are lost. Except the ghosts fought back, wrecking the infrastructure they controlled. They knew what death was.

Eventually, things calmed down. The production and presence of AI's on Earth was declared illegal on penalty of mindwipe. The existing ghosts, all that were accessable, were recalled and rebodied, with hardware limitations imposed on them to prevent rebellion. Humanity was once again top of the food chain, with no threats from their servants. 

But preventing rebellion does not prevent suicide....