Early and Late TL9

[Fred Brackin]

Originally Posted by weby

N

Artificial Intelligences

Dedicated AI: This is basically middle TL 8 tech.

Non-Volitional AI: This is basically late TL 8 tech.

.

Look at the IQ scores the UT formulas would generate. There is no computer in the world with an unspecialized broad-based/interact with the real world in real time IQ score above 2.

[weby]

Originally Posted by Fred Brackin

Look at the IQ scores the UT formulas would generate. There is no computer in the world with an unspecialized broad-based/interact with the real world in real time IQ score above 2.

Note that broad based active interaction requires Volitional AI, Non-Volitional AI will basically only respond to things, not initiate things, being describes as being an automaton, but being able to learn.

The dedicated AI is not intelligence as we humans see it. It is incapable of learning. The complexity to IQ thing is thus not same as complexity to intelligence as commonly seen. It is clearly defined as a smart tool. Thus this is a system that can substitute for a human in very narrow cases.

The Non-Volitional AI is defined as being capable of understanding speech and learning but still without initiative and so on. So this is a system that basically can substitute for a human in routine cases, but occasionally gets things wrong due to not understanding context where a similarly skilled human would understand and does not deal well with unexpected things.

First some background:

At middle TL 8 we are thus at standard HT complexity scores and in 1990 according to HT:E&E. Thus complexity 4 for a workstation.

So IQ 6 maximum in workstation for a dedicated AI.

At late TL 8 we are at 2010 or after and +2 complexity according to HT:E&E. Thus complexity 6 for a workstation or complexity 5 for a normal desktop.

So IQ 8 maximum in workstation for a Non-Volitional AI

I have selected the workstation as the level in question as that seems like a typical high end work computer or a server in a server farm.

Thus we are according to characters at:
EArlier: "6 or less: Crippling. An attribute this bad severely constrains your
lifestyle." and incidentally IQ 6 is the limit of ability to use tools and language.
and later: "7 Poor. Your limitations are immediately obvious to anyone who
meets you. This is the lowest score you can have and still pass for “able-
bodied.” or "8 or 9: Below average. Such scores are limiting, but within the human norm. "

The IQ 7 situation kind of reminds me of a lot of interactions with many "smart systems" today like the different search engines and similar.

The interesting part thus seems to be that that cannot use tools limit seems to hit at same level as I perceived the computer in question to be that level, will explain more below, but did not notice that before.

Several of our customers work in manufacturing and some of them have deployed self learning systems to help.

The first system that I would say that was a "wow" in how smart a system was a system for plasma cutting planning. The customer cut single pieces and short series of steel plates to forms customers wanted with a plasma cutter.

The basic problem was optimizing shapes to minimize scrap. Most of the pieces of steel they had already had been used, as in pieces had been cut off from them.

There were also dozens of materials with slightly different reactions to heat, with dozens of thicknesses and so on. So things like how wide the safety zones and how long the start ramp should be changed also. Too much and you lose material, too little and you have to discard a piece that does not meet specs.

So cutting the pieces on the correct piece of steel in correct orientation with the correct tolerances was not a trivial thing.

Third consideration was to not have too much material in stock to minimize the capital stuck in stock, but having less pieces in stock again meant that you were less likely to have a piece where you would lose only little extra material. And of course you have limited storage space and get some money back from scrap metal.

So they had a long term engineer doing the production and stock planning. They had at few points tried a computer system to help him place the things on the right piece of steel and so on, but had not managed to find a good tool.

That engineer then left in early/middle part of 1990s(cannot remember the exact year) and they could not find an experienced such, so they faced the prospect of very high scrap rates for a time as they trained a new person for the job. But a salesman managed to get them to try a new self optimizing system as if it worked they would save the salary.

Early on that system was about what they would expect from a new employee with significantly higher scrap rates than the experienced engineer, but nothing specially bad. But the system kept getting better and better over time and few years later it was overall marginally better than the engineer with a long experience. But every time they had a major overhaul of one of their cutters the system would be much worse for a short period until it learned the new parameters of the machines, taking much longer there than a human. Basically whereas a human would go "Ok, we had overhaul, so the tolerances are likely tighter so I need/can.." the program was just "ok, next piece" and had then to learn what worked with the new things.

As for the skill level of that program hard to say but at IQ 6 and optional specialization to a questimate Average skill it apparently had 20 points to reach the "professional skill level" of 12.

So at TL 8 HT software prices it would have cost 30 000 Gurps$ as base price and +5%/extra point=+100% or a total of 60 000 Gurps$. But in reality it cost around 12 000 about 1994 dollars or about 14 000 Gurps$ costs.

As for the late TL 8 that we are currently in there are a lot less interesting examples, but a lot more than I can point to.

Things like smart speakers/Google assistant/Siri/etc understand speech with varying degrees of success but have definitely gotten better over time and they likely get background processing power equal to a server or less for a brief period so maximally that cpx 6 for a short while.

Then we have things like the Google Game playing AI series that learned Chess, Go and other games so much better than a human, though those are usually run in higher complexity, but then you are still talking about programs that can beat the best humans.

Then we have things like the Tesla send driving, that still requires human interventions at times, but the number of interventions/distance driven has gone down dramatically in short time.

And we have all the systems producing art and many others.

The interesting part is that at least on some mobile devices the voice processing is done locally and language skill requires IQ 6 in Gurps, they have likely specialized hardware for that but still that requires Non-Volitional AI according to UT. So that would point to effective CPX of 5+ for a tiny computer at TL 8 that would indicate that 2020 we hit next CPX as expected according to HT:E&E so +3 =1(base)+3(advancement)+1(fast) for a CPX 5.

[weby]

Personal and consumer goods:

Attaché Case: likely will exist at early TL 8+1 as special security tool. Normally used attache cases will likely instead of higher DR get lighter.

Pocket Pack: a weird collection of items. I do not see why such would be sold...

Cleaning Gel: Likely possible at TL 8 and might get trendy at some point, no idea when, so up to GM.

Smart Brush:Likely possible at TL 8 and if there appears need...

Imprint Circuits: No idea why you would print such.. But clothing with Leds and other weird things exists already today.

Computer Clothing: Seems impractical as you would have a different computer with each set of clothes and would need to buy a new computer with each outfit.. So while possible today, likely computing devices would be a separate thing like a smart watch, glasses, smartphone or tablet or so on..

Varicloth: Something like this might well be popular with travelers as you need to carry less clothes with you and still have variety, specially if clothes develop to be more dirt resistant. This is something that is likely possible today or near future, but no idea on the progress of price coming down to reasonable.

Cybervox: Likely never. Likely any TL8+1 multi-function musical device like that will continue to be a peripheral to a smartphone/tablet/laptap where the actual thing happens and the device is just an additional interface..

Euphoria Machine / Ecstasy Machine / Neurostimulator: I do not believe in near future neural interfaces to be widely used, will talk more about those in medical part. The non neural interface part seems.. questionable at least any time soon. So any of those might came at late TL 8+1, later or never.

Android: As defined by UT is a humanoid robot that may or may not be realistic human looking. I think those will come into wide use in early TL 8+1. The reason that a humanoid robot is beneficial is primarily that our world is built to be used and operated by humans. I am talking about everything from stairs to controls for things. So most things that would need to be operated by a nonhumanoid robots would require some modification for that, whereas a humanoid robot can use them directly. The schedule of adaptation will depend on how fast we get GAI. But I would guess that by early 2030s we will start seeing a lot of them and by late 2030s they have taken over most menial jobs and some not so menial unless there is strong intervention to stop it, but that is just an educated guess.

Petbot:I think these will come at very early TL 8+1 and become very popular as the required usefulness is multiple degrees lower than an android.

[Astromancer]

The first house cleaning robots to be more than vaguely stooped up Rumbas will be in places like hospitals and industrial clean rooms. Because that is where they will make sense.

As nursing staff becomes rarer and more expensive, robot assistants will make more sense. Lifting patients out of bed causes vast numbers of injuries among nurses. They already have prototype nursing assistant robots in Japan.

Because of the constant need for exacting cleaning hospitals actually are demanding places for janitors to work. Automated cleaning makes practical sense here. Later as the technology improves, it will get cheaper. At some point upper middle class households ( upper class households hire servants) will get cleaning robots.

The average home having a cleaning robot is late TL 9.

[Fred Brackin]

Originally Posted by weby

P

Pocket Pack: a weird collection of items. I do not see why such would be sold...

C

It's for spacers and the "pocket" it goes in is most often on a spacesuit or a survival suit. The item dates from Space 1e back in the 80s.

[Inky]

Yeah, it seems to be basically just a "mini survival kit" of the kind you see various variations of in camping shops and novelty mail order catalogues - just with the components upgraded to TL9, like the food tablets and the "vacuum-proof" duct tape. Useful to grab as a back-up policy if you're doing your character's equipment in a hurry.

[Willy]

I think we allways should keep in mind that some tech things are already developped and ready for common use, but still the old tech is filling their niche.

A exampel is transport we could have malev trains and subwys, but why should we build them actually, we have invested billions in a more or less working system of railways and subways that fit to our needs. Building them from scratch ist near impossible, just consider the work digging new tunnels in existiung citys.

Same goes for a lot of other fields why buy and use new inventions if the old stuff is reliable and good enough, not to mention the old tech producers want to make more money with the old stuff and play stick in the mud.

Therefore in history often totally wrecked down places like post WWII germany or the balticum after the fall of the USSR started fresh. Building the newest and best they could get and having afterwards a significant advantage over places who still have the old tech. German industry had the most modern factories of post WWII and the balticum has a completely new Internet setting a standard for most of europe, because of this. Germany in the opposite has the old reliable net with much slower speed and other problems.

If we discuss this topics we should always keep in mind which way would we go if we had to build a nation from scratch using what is available now.

[mlangsdorf]

Originally Posted by weby

Mag-Lev Subway and commuter trains: Likely not in early TL 8+1 at least as the trials have not caused any real world interest.

Assuming there are cheap-as-steel-rails room temperature superconductors at early TL 8+1, I could see maglev bullet trains replacing existing bullet train lines. The economics don't work for subway/commuter lines: no friction with the rail is better when you want to go very fast, but commuter lines are stopping and starting a lot and can't go very fast anyway.

Originally Posted by weby

Ballistic Liners: This one is more difficult, as due to noise and such things Supersonic transportation is hard to make widely available...

Air Car : This is one of those.. Basically because of the need to move a lot of air causes and lot of sound...

Tilt-Rotor Transport/Tactical Tilt-Rotor: The technology exists at late TL 8 and is fully viable, but problematic in that as you need to make the proptors smaller than helicopter rotors they need to spin at higher RPM...

Utility Vertol: Same problems as the air car described earlier...

Assuming vortex ring projectors are early TL 8+1, that means TL 8+1 went through a fundamental transformation in understanding fluid dynamics. The same technological underpinning that lets a 2 lb pistol deliver a cohesive sphere of smoke at 15 yards probably lets you do a lot of weird things (from a TL 8 perspective) with sound and propwash. If the crack of a supersonic airliner dissipates into generalized rumbling after traveling 25 yards, or the propwash of a tiltrotor, aircar, or utility vertol is minimal, then those vehicles would probably be more appealing.

Arguably, vortex ring projectors are superscience. But if they're not, that technology (or variants of it) should show up in a lot of places that care about airflow.

[Fred Brackin]

Originally Posted by mlangsdorf

A

Arguably, vortex ring projectors are superscience. But if they're not, that technology (or variants of it) should show up in a lot of places that care about airflow.

Vortex ring projectors are probably small and unimpressive superscience.

"You violated the Laws of Physics for _that_?"

The revolutionary/not understood at our current level propulsion thing probably has a name like 'Tesla Bladeless Turbine" (which term might actually get you some google hits). It might take you to gernsback in the IW book too.

[Rupert]

Originally Posted by Astromancer

Because of the constant need for exacting cleaning hospitals actually are demanding places for janitors to work. Automated cleaning makes practical sense here. Later as the technology improves, it will get cheaper. At some point upper middle class households ( upper class households hire servants) will get cleaning robots.

The average home having a cleaning robot is late TL 9.

The other reason cleaning robots will be first used in industrial clean rooms and hospitals is that they have simple, orderly, and consistent layouts. Cleaning a home or a heavily used office is not something a robot will be good at for quite some time. Vacuuming the floor is something they will manage sooner, but the rest of the job will take longer before they're up to it. Oh, and cleaning toilets should be fairly simple, but for institutional toilets it'll be easier to make them watertight and have them hose themselves out every so often.