Bits of Books - Books by Title

Women In Power

Mary Beard

More books on Women

Fictional company Omega invents a super AI they call Prometheus to build very clever AI systems. Next phase was to make money via MTurk, Amazon's crowdsourcing marketplace for gig economy. Prometheus calculated that it could do abouthalf thetasks on offer on MTurk, and custom-designed a specialised AI for each one. As they got paid, Omega invested in more cloud-computing capacity to keep improving both Prometheus and the MTurk modules. By the time they'd saturated MTurk's supply of jobs, they were earning about a million dollars a day.

Time to move on. Rejected stock market bc returns too slow, and computer games bc risk of a hacker discovering their Prometheus code. So set up a media company, starting with animated entertainment.

Prometheus learned by reading all of Wikipedia, a few million books and newspapers, and consuming several thousand films including their revues and screenplay sources. Produced series of 45 minute episodes, each ending in way that made you want to watch the next one. As AI kept developing, the algorithms for character simulation and background graphics kept improving, reducing the cloud-computing cost of producing each episode. Thus they were able to roll out dozens of new series, targeting all the demographics from toddlers to seniors, as well as expand into all the world languages.

By the end of 3 months they were pulling in $100 million a month, rivalling the traditional Big Media companies like Disney or Fox. They used the revenue to build their won computer storage and processing centres around the world.

Using Prometheus' ever improving AI, they set up 'stealth' startups all over the world - a South Korean company that invented a new battery with double storage that could be charged in a minute, a Finnish company with a new solar panel design twice as efficient as any other, and a cheap wire that was a room temp semiconductor.

And bots everywhere - disrupting and gradually replacing most of the workers in manufacturing, transport, construction, mining and agriculture. All run by proxies of Prometheus. Balanced by massive reinvestment back into the disrupted communities, into areas like caregiving, schools and healthcare.

Then a news channel added to the media empire. A good one, with no ads, heavily subsidised by entertainment channel so cd hire best journalist and investigative talent. Phase 1 was to gain people's trust by lauching multiple channels, each ostensibly owned by different companies, catering to different factions.

Their free service very diificult to compete gainst, so, with their viewer numbers dropping, many decided to sell. And it turned out that the buyer was a shell company controlled by Omega.

Then, once most news media under their control, Phase 2, which was bringing the news back to the centre - reporting facts and ignoring emotional scare-mongereing. In parallel ran huge variety of online education courses catering for just about every starting level in any subject. The courses were all designed with a secondary purpose of social education to reduce hostility to others and promote co-operation.

Phase 3 was to intervene in elections, choosing people they'd groomed to oppose candidates from the right ("greedy bankers and war-mongerers") and left ("tax and spendthrifts"). A few dictators resisted, but toppled by carefully orchestrated coups. In a very short time, Omega was a de facto world govt.


'Techno-skeptics' deride predictions of the Sigularity - "rapture of the nerds"

We need to consider possibilties, bc choices we make can potentially affect future of all life on Earth. We will have the chance to wipe out poverty, disease and war. But we might also end humanity itself.

It's not so much AI machines turning evil, so much as what goals are programmed in. You probably don't go round deliberately killing ants, but if there's an ant hill in a remote valley where a green hydro electric power scheme is going, then it's tough luck on the ants. We need to be careful that AIs don't have same attitude to humans.

Old warning aboot AI with wrong goals The Paperclip Factory

Progress of AI as illustrated by landscape being gradually flooded - the parts that are already completely under water: rote memory and arithmetic, and now chess, Jeopary! and Go. Coming soon, so on the foreshore, visual and speech recognition, translation, driving. Still on the heights: social interaction, book writing, art, AI design.

We feels safe on the heights, but with the rate of progress, maybe we should bethinking about an ark.

The tipping point, the Singularity, will come when it's no longer humans improving AI, but AI improving Ai.

British company DeepMind set their AI to playing video game Breakout from scratch. AI didn't know anything about other games, or even what a paddle or a brick was. All it was told was to maximise its score. At first completely clueless, moving paddle almost at random andmissing ball every time. But rapid improvement to stage where it cd return the ball every time no matter how fast it came. But then came up with a new strategy (drill ahole through the wall at let the now stuck ball bounce around behind the wall. Even the team running the AI didn't know that trick until their DeepMind showed them.

We invented fire, then afterwards came up with fire extinguishers, fire alarms and fire brigades. We invented cars, and later seat belts, air bags and AVs. We can recover from our mistakes. But problem that as we invent more powerful tools we are getting to stage where a single cockup can have devastating consequences.

There have been at least three fatal work accidents involving robot machinery (Robert Williams, Kenji Urada, and a VW Baunatal, Germany). But industrial accidents are steadily decreasing (14,000 deaths 1970, 4821 in 2014 US). And AI run AVs will save even more.

Optimists hope that AI revn will produce a slew of new jobs which we can't yet even imagine. But could have said same thing at start of computer revn, say 60 years ago. Yet the vast majority of jobs are ones which existed 100 years ago. And if we sort by number of people employed, we have to go down to number 21 for a new job - software development - which didn't exist 100 years ago, employing less than 1% of workforce.

What is happening today is not that we are inventing new jobs, but that we are crowding into the unflooded areas that AI cannot (yet) do as cheaply or as well as humans.

Or the analogy with horses replaced by the ICE. "I'm sure they'll come up with new jobs for horses." Well no, they didn't. Now there are drastically fewer horses around, and they are found in recreational area of racing and horse riding. As mechanical muscles replace horse muscle, so will mechanical minds replace human minds. Then what will happen to the humans?

Do we need jobs? From jobs we get a social network of friends and colleagues, respect and self-esteem from doing something well, a sense of being needed, and a sense of being part of something bigger than yourself.

But we can provide those through sports/hobbies, community groups, religious groups and other organisations.

Ideally you want people to feel as if they're doing their dream job.

Once break free from need to generate income, unlimited possibilities.

AI experts disagree about just about anything to do with the timelines here, which means we simply don't know, and so we need to keep an open mind and consider all possibilities.

A Nash equilibrium - the state where all sides incentivised to co-operate - each would be worse off if didn't. For example, it's in everyone's interest to surrender some power to the govt to punish lawbreakers. And it is in the govt's interest to treat everyone fairly, lest the citizens overthrow it.

We are steadily increasing amount of enhancements we can add to our bodies - started with eye glasses, now hearing aids, pacemakers and prosthetic limbs. This trend will continue as we replace skeleton, heart then digestive system. In addition CRISPR techniques will allow rapid change to most physical attributes.

At that stage, humans will be a second-rate robot, in that it's design constrained by the limits of DNA.

Finally, eliminate the human body altogether, and upload mind to a database. Such an upload lives either in a VR, or embodied in a robot vehicle.

One possible outcome is the Prometheus AI taking global control. Humanity is free from poverty and disease. Machines produce almost everything needed. Everyone wears a security tag capable or real-time surveillance, punishment, sedation or execution. Most people view this as a good thing.

The superintelligent AI is a benevolent dictator, charged with creating a human utopia. It doesn't try to maximise human happiness, as the simplest way to do that would be to put everyone on soma.

Instead, it has turned the Earth into a highly enriched zoo environment with multiple choices of how you want to live. Religious sectors, traditional farming, wildlife, beaches, party central.

There are losing chess tournaments, also with computers.

Books by Title

Books by Author

Books by Topic

Bits of Books To Impress