What could possibly go wrong?

[valhalla360]

Originally Posted by s/v Jedi

One more observation: there’s talk of robots and AI and then there are remarks like “a hall of workers is now a hall of robots”. It seems there is a misunderstanding of what AI is. The industrial robots are just computerized tools, there is no AI for the vast majority of it. It’s like the lathe that can cut a thread, not a machine that is intelligent.

Yes & No.

As I said, lots of people see AI and think Data from Star Trek. Most AI is in the electronics not in the physical device. Reaching that level of AI is likely beyond our lifetime.

While industrial robots are mostly automation, they are moving toward AI.

40yr ago, a robot to put a screw into a product needed to have the screw arrive at a very specific location in a very specific orientation. The product needed to be exactly where it was supposed to be, again in a very specific position and orientation. This made them really only viable for mass production because it takes a huge amount of human effort to design it and work the bugs out of how to make that happen. Where you are building thousands units a day, automation is still likely to rule the day over AI....but as it becomes better and easier to implement, expect it to creep in. If the robot can look at the screw and think about the orientation and adjust how it picks it up, a lot of physical feeder systems for aligning the available screws can be eliminated. Makes it much easier when the engineers spec a different size screw that doesn't fit in the old screw feeder. Now instead of a $50,000 screw feeder system, you can put a simple plastic bin of screws next to the robot and it figures out how to pick them up.

There are lots of robotic tasks that require (or benefit from AI). Take an Amazon warehouse with automated stocking and shipping robots. As products are constantly changing, orders can be filled faster and with less wear & tear on the robots if commonly purchased items are stocked closer to the loading docks. If the robots can decide which new products are likely to sell fast, they can save the company money by stocking those nearby. This is a perfect use for AI.

How about restaurant apps where it looks at feedback and adjusts how the food is prepared. If you post a review that the pizza was too saucy and a little doughy, it could automatically reduce the sauce by 20% and cook it for 1:35 longer. It could also make more subtle adjustments for customers who fit a similar profile as you who didn't make review comments. In the end dialing it in to produce a product that better fits the customers desires.

[valhalla360]

Originally Posted by s/v Jedi

I mentioned an “R2” before but it got lost in the data Instead of looking at Data, look at a simple R2 like R2D2 which is a droid to fix problems aboard the craft like X-Wing fighters but also utilized for countless other functions. It’s something close to what we’re able to achieve.

In that respect, Star Wars likely gets Androids better than Star Trek.

C3P0 would probably be able to make you dinner but a specialized cooking android would be better suited to the task.

But that's not much different from humans. What NFL linemen are 120lb vs 300lb guys? It's not that a 120lb guy can't grasp the principals of blocking (the AI portion), he just isn't physically suited to the task.

[belizesailor]

Originally Posted by s/v Jedi

But there are more differences now than what we had in the past: where this hi tech upper class used to be easily able to suppress the “primitives”, this is not the case anymore: the primitives not only got the same guns but they are better at handling them than this upper class… which undoubtedly is scheming to send in killer robots etc. The movies show it all.

Also, there are many in this upper class who choose to switch over to the other side. This too has happened in the past (like the Rockefeller son becoming a tribe member in Papua New Guinea iirc) but is much more prevalent today.

The mix is certainly much more complex today and for the first time tech enables it to be instant and global.

Interesting times a'comin!

[belizesailor]

Originally Posted by valhalla360

In that respect, Star Wars likely gets Androids better than Star Trek.



C3P0 would probably be able to make you dinner but a specialized cooking android would be better suited to the task.



But that's not much different from humans. What NFL linemen are 120lb vs 300lb guys? It's not that a 120lb guy can't grasp the principals of blocking (the AI portion), he just isn't physically suited to the task.

At least in terms of robotics, look at the amazing advances by companies like Boston Dynamics just in recent years. It seems pretty obvious that robots like theirs will have industrial & military applications. Making a specific purpose (not full on AI) device like R2D2 is even in the reach of hobbists now.

https://www.google.com/url?sa=t&sour...JWX6_boAfF9vpS

The AI part for a CP30 like creation is much further out, but if past tech change is an indicator, I expect the development curve on that too will shorten dramatically soon.

It seems highly likely to me to that we will see a cyborg-ish path for humans. Just look at the great advances in prosthesis devices in recent years. Its not much of a stretch to imagine a robotics/prosthesis/human convergence.

Some might interpret the cyborgish development of humans to be a horrid abomination, but in practical terms think how dramatically that can improve the lives of those who are injured or disabled.

The future Pirate Captain, or disabled cruiser, will likely not have a peg leg or a hand hook, but a much more functional cyborg one. And thus be able to sail much better!


https://youtu.be/Gl4ghrrXmRo

[valhalla360]

Originally Posted by belizesailor

At least in terms of robotics, look at the amazing advances by companies like Boston Dynamics just in recent years. It seems pretty obvious that robots like theirs will have industrial & military applications. Making a specific purpose (not full on AI) device like R2D2 is even in the reach of hobbists now.

The AI part for a CP30 like creation is much further out, but if past tech change is an indicator, I expect the development curve on that too will shorten dramatically soon.

In many ways the AI part of CP30 is already here. A basic smart phone can translate most languages. The only reason it can't speak Hutt (as in Jabba), is because it hasn't had a chance to learn it. Use the camera and it can translate text on books, signs or menus. Want random trivia, it will search the web and provide it. If you turn on the full functionality, the phone will try to determine what you want and join conversations without even being asked.

R2D2, on the other hand, trying to diagnose mechanical issues on the fly, formulate a solution and then implement it...is further away.

Of course, the implication was never that R2D2 wasn't as intelligent just incapable of speech...which given the speech capabilities of modern electronics no longer makes sense, since it's such a basic functionality.

[valhalla360]

Another thing people confuse with AI is they see an individual failure and think that is proof that AI doesn't work or won't happen. Such as this boat turning around due to a mechanical issue.

In reality, it's the way most AI operates. It's a lot like evolution. It tries lots of variations. Many (often most) will fail. Sometimes, it can do it via simulations rather than actually undertaking the task. That's a big part of how it learns. After thousands of transatlantic crossings, it tries to avoid the choices where the boat sank, arrived at the wrong destination or just took a really long time. It concentrates on repeating the ones where the boat arrives quickly and safely. The system is constantly learning and optimizing.

Human intelligence often appears random because there are so many inputs and so many desired outcomes... It becomes difficult to follow the logic compared to a simpler AI task.

[ArmyDaveNY]

Originally Posted by valhalla360

This is the biggest thing slowing things down. There have been various automated driving systems for several decades. Yes, the newest versions are better and able to handle more situations but there have been significant benefits missed as a result.

But...
When an idiot causes a crash, they shrug and move on.
When an automated car causes a crash, the lawyers start drooling over the multibillion dollar company they can liquidate in a class action lawsuit. As a result, major companies are hesitant to be the first to market.

Indeed! I am, to an extent, surprised that some legislature somewhere hasn't adjusted their laws to limit liability so as to attract manufactures and developers. I suspect that at some point someone will figure this out and then there will be an increase in the rate of deployment.

Given the amount of sensors and the service they will require, I would not be surprised if companies will only lease you an autonomous car or boat at some point in the future. What company will allow someone to drive for more than three (four, five?) years with no assurance that the driver will regularly and properly service the systems needed to function properly.

For businesses, they may sell their products but they will almost certainly require a concurrent service contract that ensures the technicians properly take care of the system.

It is also possible that the technology will change/improve so quickly that they manufacturers may develop universal installations so that the hardware required to support the software can be swapped out with little effort or time required.

As Mr. Dylan said, "..the times, they are a-changin'."

[CaptTom]

Anyone have any news on the "AI Mayflower?"

It's now showing 10% battery power, no propulsion and 1 knot speed.

Their web site seems to be big on flashy marketing, and short on details.

Their Twitter feed is just plain goofy. Again, no details.

Honestly, I don't know how they can get people excited about a project they won't even talk about!

[waterman46]

Originally Posted by NaClyDog

I don't think this is a fair or accurate statement.

You will still need humans... yes, they may be doing something different but humans are still required for automated whatevers.

But wait a minute, the press release said this vessel will be "thinking"! Who the H needs humans anymore?

Really my irritation with this project is that they presume to call it some version of "Mayflower". What an insult to the memory of the brave souls who struggled across the sea in search of an unknown future in a journey that today would be considered impossibly risky and illegal (passenger service in a vessel woefully un-equipped for safe navigation and possibly uninspected by any authority).

In fact this sort of remote operation should not be allowed to take place. International rules require a proper lookout to be maintained at all times. To maintain a lookout remotely would be impossible without relying on a chain of electronic devices, any one of which failing would cause failure of the lookout.

[hpeer]

Little Wing,

Based in your recommendation I bought and read Shoo Class. Good read; interesting observations. The guy has a very interesting life story.

Thanks for the suggestion.

[CaptTom]

Their tracker site shows the vessel back in port.

Still no update on the web site or Twitter. I've totally lost any interest I might have had in this project. They wanted to grab the headlines when things were going well, but at the first sign of trouble they go silent.

[s/v Jedi]

Originally Posted by CaptTom

Their tracker site shows the vessel back in port.

Still no update on the web site or Twitter. I've totally lost any interest I might have had in this project. They wanted to grab the headlines when things were going well, but at the first sign of trouble they go silent.

The whole AI on Twitter thing is childish and must have come from some marketeer twisted thoughts. It also completely ridicules serious AI applications.

[LittleWing77]

Originally Posted by hpeer

Little Wing,

Based in your recommendation I bought and read Shop Class: https://www.amazon.com/Shop-Class-So...p+class&sr=8-1Good read; interesting observations. The guy has a very interesting life story.

Thanks for the suggestion.

Glad you liked it! I re-read it about once a year. He's written another couple of books more recently. I'm still trying to get through The World Beyond Your Head https://www.amazon.com/gp/aw/d/03745..._dp_0374535914

His most recent offering is Why We Drive: Philosophy of the Open Road - https://www.amazon.com/gp/aw/d/00627..._dp_0062741969

Maybe I should switch to that and I'll have better luck!

Overall though, I think we just need to get Matthew on a sailboat...
Fair winds,
LittleWing77

(Sorry for all the Amazon links. A bunch of Matt's books are also much more affordably available through thriftbooks.com and abebooks.com. I was being a bit lazy, because I consciously choose to use Amazon as seldom as possible.)

[LittleWing77]

Originally Posted by CaptTom

Anyone have any news on the "AI Mayflower?"

It's now showing 10% battery power, no propulsion and 1 knot speed.

Originally Posted by StuM

If you're a Twitterer:
https://twitter.com/AI_Mayflower

"Mission update: I am safe and I still have power. Awaiting recovery - support team on way. Thanks for all your messages of support!"

This is (one of) the points made most succinctly by Bongo in Post #27, CaptTom.

The MAS has had an operational failure and is sitting there, effectively Not Under Command (NUC), posing a hazard to navigation until a team of HUMANS can get there and hopefully sort things out.

How many days has it been? How long will it take to fix at sea?
LW77

EDIT:

Originally Posted by CaptTom

Their tracker site shows the vessel back in port.

... They wanted to grab the headlines when things were going well, but at the first sign of trouble they go silent.

So they've got her back to port with their tails between their legs. Of course, there's no fanfare! What an epic fail.

[RiverRat37]

Originally Posted by valhalla360

What a lot of people miss is there is a spectrum of AI. If you think of it in terms of organic lifeforms:


Depending on the system and goals, we are probably somewhere on the level of insect-lizard but it's advancing fast.

I sincerely doubt we are anywhere near an insect or lizard with AI development. They've been around for millennia ... successfully. That takes a lot more than you're giving them credit for. Just sayin'.

Oh, and "advancing fast" is also on a spectrum. Compared to where we hope to be 'soon', I'd say were all the way up to 3mph (+/- 1mph). We'll need to be at speed of light to get where people believe AI is going.