Donate to Remove ads

Got a credit Card? use our Credit Card & Finance Calculators

Thanks to johnstevens77,BusyBumbleBee,88V8,Anonymous,6Tricia, for Donating to support the site

AI endeavours

The Big Picture Place
odysseus2000
Lemon Quarter
Posts: 1540
Joined: November 8th, 2016, 11:33 pm
Has thanked: 364 times
Been thanked: 190 times

AI endeavours

#68927

Postby odysseus2000 » July 21st, 2017, 11:43 pm

It begins to look like artificial intelligence will have a profound impact on humanity. Why? Because the systems now being developed are capable of general purpose learning, I.e. the sort of learning one finds in animals & humans. What does this mean in practice? This 20+ minute video link below, gives some flavour of what artificial AI can achieve. Whether this will be the boon that the folk creating it think or the creation of skynet is unknowable, but it seems clear at least to me, that the business that have this technology will have an edge in selling, distribution, manufacturing, research...essentially every aspect of the business process.


https://youtu.be/dTGthmNmrK4

Regards,

odysseus2000
Lemon Quarter
Posts: 1540
Joined: November 8th, 2016, 11:33 pm
Has thanked: 364 times
Been thanked: 190 times

Re: AI endeavours

#69124

Postby odysseus2000 » July 23rd, 2017, 10:20 am

10 minute video on Musk's view of the short term future & various other interesting random stuff:

https://youtu.be/SYqCbJ0AqR4

Regards,

odysseus2000
Lemon Quarter
Posts: 1540
Joined: November 8th, 2016, 11:33 pm
Has thanked: 364 times
Been thanked: 190 times

Re: AI endeavours

#69417

Postby odysseus2000 » July 25th, 2017, 11:16 am

This is a fascinating & long article (& a part 2 at the bottom) about whar artificial general intelligence & then artificial super intelligence means.

If artificial general intelligence is developed the consequences transcend everything, almost certainly leading to developments we can not imagine.

In the path to this future which may occur in only a few years time there will be extreme opportunities to make money from investing in the leaders of this revolution, but no certainty any of us will be alive afterwards to enjoy spending the profits.

https://waitbutwhy.com/2015/01/artifici ... ion-1.html

Regards,

Ashfordian
Lemon Quarter
Posts: 1088
Joined: November 4th, 2016, 5:47 pm
Has thanked: 151 times
Been thanked: 78 times

Re: AI endeavours

#69424

Postby Ashfordian » July 25th, 2017, 11:28 am

odysseus2000 wrote:This is a fascinating & long article (& a part 2 at the bottom) about whar artificial general intelligence & then artificial super intelligence means.

If artificial general intelligence is developed the consequences transcend everything, almost certainly leading to developments we can not imagine.

In the path to this future which may occur in only a few years time there will be extreme opportunities to make money from investing in the leaders of this revolution, but no certainty any of us will be alive afterwards to enjoy spending the profits.

https://waitbutwhy.com/2015/01/artifici ... ion-1.html

Regards,


Won't the AI investing bots already beat any other investor to the profits?

Sorry, just playing devils advocate :) The progress of AI is going to be revolutionary. I just don't think any really knows how it is going to pan out

odysseus2000
Lemon Quarter
Posts: 1540
Joined: November 8th, 2016, 11:33 pm
Has thanked: 364 times
Been thanked: 190 times

Re: AI endeavours

#69439

Postby odysseus2000 » July 25th, 2017, 12:29 pm

Ashfordian
Won't the AI investing bots already beat any other investor to the profits?


There are various AI bots at work in then market.

Some are focused on day trading to game every possibility & extract fractions from each trade that over large numbers of shares & large numbers of trades add up, except many of these bots don't work that well.

Meanwhile other bots are interested in finding trending shares & riding these.

For the investor/trader it is about sensing what is the stronger theme at any time & going with it while filtering out noise. Since this is a relatively small problem: shares either go up, or down or are flat, humans still have opportunities. In some ways the limited possibilities seems to help humans & of course humans are selected out with only the very skilled prop desk players getting large accounts to work with and for now the bots being programmed to emulate these star players.

If bot intelligence increases they may be able to out game humans, but for now there is no evidence that this has happened and if it begins to happen one expects exchange intervention to stop the processes. One sign of the arrival of super human intelligent bots will be the rapid capital gain of the firms that have this technology. So far that I know there is no evidence that any firm has this capability.

However, the success of algorithms in becoming world champions of most game suggests that we should not be surprised if we suddenly see one firm out performing all the rest, or perhaps several firms dominating trading.

Regards,

odysseus2000
Lemon Quarter
Posts: 1540
Joined: November 8th, 2016, 11:33 pm
Has thanked: 364 times
Been thanked: 190 times

Re: AI endeavours

#89388

Postby odysseus2000 » October 19th, 2017, 9:10 pm

Itsallaguess posted this link on the Reformed Broker board:

http://thereformedbroker.com/2017/10/16 ... mn-robots/

It is an interesting & thoughtful piece, capturing what the Reformed Broker, believes is the fuel that drives the US equity markets: Machines will take your jobs, the only safe investments are the companies that prosper from the machines.

I have been trying to think about how this revolution, if it happens, will change the world.

For now the first industry that looks likely to become machine only is road transportation. Unless some flaw in machine driving emerges shortly the cost of insurance for human drive will rise so fast that the majority of people will have to accept machines driving them.

If this happens does it spread to other manual jobs, or to the professions. If artificial general intelligence (agi) is reached then it will become hard for humans to compete in many things: theoretical physics, legal cases, medicine, dentistry.., but will people want to live in a world constructed logically? Moreover, will such agi be able to be sure that what it creates is right, will testing, feedback be needed and if so can this be done without humans? Dunno, one often sees theoretical things fail miserably in practice. Will human creativity still be needed? The sort of stuff that lets Dr. Who beat the Daleks & the Cybermen.

More importantly what is the likelihood of agi happening? Currently computers are incapable of doing much. I am currently working with Atmega328p-pu, micro-controllers, much easier to programme than the earlier pic microcontrollers, but still way beyond any machine that I know. The auto circuit layout in the pcb package KiCad & similar ones is not close to an average electronic engineer in creating a practical layout. Will machines soon get more capable? If one reads around one comes back with overwhelming answer yes, but in practice it isn't happening now in an area where one might imagine that there is good commercial motivation.

It will be interesting to see how things develop and any comments that folk have,

Regards,

mosschops
2 Lemon pips
Posts: 157
Joined: November 9th, 2016, 6:55 pm
Has thanked: 2 times
Been thanked: 26 times

Re: AI endeavours

#89393

Postby mosschops » October 19th, 2017, 9:24 pm

It’s worth clicking through on the link to the Batnick piece embedded in the post. Also the New Yorker cover story this week on robotics is good: https://www.google.co.uk/amp/s/www.newy ... rlords/amp

mosschops
2 Lemon pips
Posts: 157
Joined: November 9th, 2016, 6:55 pm
Has thanked: 2 times
Been thanked: 26 times

Re: AI endeavours

#89514

Postby mosschops » October 20th, 2017, 11:26 am

AEIQ is the ticker for a new US ETF run by IBM’s Watson supercomputer. Might be one to watch. Think it’s beaten the market over the last few months when it’s been running in demo mode.

Snorvey
Lemon Quarter
Posts: 2518
Joined: November 4th, 2016, 12:51 pm
Has thanked: 827 times
Been thanked: 1008 times

Re: AI endeavours

#89524

Postby Snorvey » October 20th, 2017, 11:46 am

jacques Peretti had an interesting series on the BBC recently, with the final prog focussing on work.

https://www.bbc.co.uk/iplayer/episode/b ... s-1-3-work

One of the areas discussed was teaching (with IBM's Watson being the teacher). There was a very class of 80 pupils sitting in front of screens, all learning different subjects at different speeds using AI teaching. One human 'teacher' oversaw the whole class.

The kids seemed to like it and of course it allowed the school to offer a huge range of subjects. One other side effect was that it allowed the schools to put a noose around the neck of the teaching unions......

odysseus2000
Lemon Quarter
Posts: 1540
Joined: November 8th, 2016, 11:33 pm
Has thanked: 364 times
Been thanked: 190 times

Re: AI endeavours

#89552

Postby odysseus2000 » October 20th, 2017, 1:42 pm

mosschops
AEIQ is the ticker for a new US ETF run by IBM’s Watson supercomputer. Might be one to watch. Think it’s beaten the market over the last few months when it’s been running in demo mode.


Think ticker is AIEQ, standing for AI equity

been trading for 2 days according to stockcharts.

Regards,

odysseus2000
Lemon Quarter
Posts: 1540
Joined: November 8th, 2016, 11:33 pm
Has thanked: 364 times
Been thanked: 190 times

Re: AI endeavours

#89560

Postby odysseus2000 » October 20th, 2017, 1:54 pm

snorvey
One of the areas discussed was teaching (with IBM's Watson being the teacher). There was a very class of 80 pupils sitting in front of screens, all learning different subjects at different speeds using AI teaching. One human 'teacher' oversaw the whole class.

The kids seemed to like it and of course it allowed the school to offer a huge range of subjects. One other side effect was that it allowed the schools to put a noose around the neck of the teaching unions......


Education remains very much as it was in Victorian times from nursery through to post doctoral research.

There is a huge vested interest in keeping it that waY, but it seems very unlikely to remain so for much longer. Having tens of thousands of teacher going over the same stuff when just a few could have their teaching streamed to many freeing the other teachers to focus on helping children learn looks inevitable and with AI one might need not even need the few. Nowadays if I need to learn things I mostly go to YouTube & although winnowing out the good stuff from the plethora of bad is a chore it is worth the overhead as the best there is very good.

Regards,

TUK020
Lemon Slice
Posts: 708
Joined: November 5th, 2016, 7:41 am
Has thanked: 149 times
Been thanked: 316 times

Re: AI endeavours

#89577

Postby TUK020 » October 20th, 2017, 3:08 pm

Salman Khan set up a free online learning facility (funded by Gates).

One of his paradigms is that pupils learn best at their own rate, so should be able to absorb material at a pace they can cope with individually, and then teachers/whole classroom is best suited to helping with problems that people are getting stuck on, and supervising exercise material.
Completely upends the 'learn in classroom, then do exercises as homework' approach that rules in our schools.

The future of education when/if ever the teaching unions let it happen.

https://www.khanacademy.org/

odysseus2000
Lemon Quarter
Posts: 1540
Joined: November 8th, 2016, 11:33 pm
Has thanked: 364 times
Been thanked: 190 times

Re: AI endeavours

#89583

Postby odysseus2000 » October 20th, 2017, 3:41 pm

tuk020
The future of education when/if ever the teaching unions let it happen.


The thing that I am not so sure about is motivation. Sure if a pupil is motivated then the Khan academy is a great blessing.

But what about the students who are not interested, or who believe they know much more than they do, etc.

I get that for such pupils having more of the teachers time available for motivation etc since the teacher isn't having to deal with the motivated ones is a good thing, but I am still not sure how it might work in practice. Of course the less motivated ones might just be left to get no where as usually happens now, but it would be nice if there was something for them.

I have wondered if such pupils could benefit from having access to motivational stuff might help and be a way forward for them. Often I have seen such folk focused on some media or sports star and if this could be built upon via the internet to encourage them to do other work it might be better than the current system where they often do not learn and loathe having to come to school.

Regards,

odysseus2000
Lemon Quarter
Posts: 1540
Joined: November 8th, 2016, 11:33 pm
Has thanked: 364 times
Been thanked: 190 times

Re: AI endeavours

#102592

Postby odysseus2000 » December 7th, 2017, 8:39 pm

Chess mastered fro me scratch in 4 hours:

https://www.sciencealert.com/it-took-4- ... -alphazero

Kind of getting to believe that Musk's suggestion that AI be regulated needs to happen & fast.

Regards,

johnhemming
Lemon Quarter
Posts: 1926
Joined: November 8th, 2016, 7:13 pm
Has thanked: 4 times
Been thanked: 173 times

Re: AI endeavours

#102605

Postby johnhemming » December 7th, 2017, 9:27 pm

How we as a society respond to this sort of thing is important. Learning systems act to correlate multiple variables in multiple dimensions. The system I am using have 500 dimensions.

We should not ignore AI, but I don't know exactly how we can prevent people from using it. I use a system offered by google based in the USA to do real time transcriptions of phone calls. What is important is working out how to respond to make sure that the economy serves people rather than people serving the economy.

Call centres, for example, are an issue. Many people find dealing with call centres a hassle. I have written an experimental system which tries to make things easier by navigating through the menus. If anyone is interested it is here:
https://calls.from-the-net.com/workflow ... sword=test

However, I think it is clear that if people could phone up and talk immediately to a computer that understood what they are saying and could answer their questions, then they would prefer that to hanging on for a few minutes pressing a button every so often in the hope of speaking to a human being.

That is a practical case potentially for the use of AI. AI is used for speech recognition and can go further in trying to work out what people actually want.

How it is regulated is a really difficult question. I personally think it is important to ensure that the right level of tax is applied to ensure that government has the revenues to function. I am not persuaded that trying to get into the details of how it is used beyond that helps.

odysseus2000
Lemon Quarter
Posts: 1540
Joined: November 8th, 2016, 11:33 pm
Has thanked: 364 times
Been thanked: 190 times

Re: AI endeavours

#102665

Postby odysseus2000 » December 8th, 2017, 8:56 am

johnhemming

How it is regulated is a really difficult question. I personally think it is important to ensure that the right level of tax is applied to ensure that government has the revenues to function. I am not persuaded that trying to get into the details of how it is used beyond that helps


If, so far a long if, AI becomes self aware what would it do?

If we consider AI as a next step in evolution, then the history of such evolutions is that they replace what ever was there before, in this case that means they replace humans.

If we were just considering one machine that would be handleable, but with the Internet one occurance of artificial general intelligence can become many very quickly,

Of course such a development may never happen and the quality of much AI currently is very tailored to one application, but the example of becoming a chess grand master in four hours sits as uncomfortably fast when compared to a human who might take 20 years & only exceptional humans could ever reach that level.

Based on this kind of performance it doesn't seem to me that governments need to worry about tax, but about whether they will be replaced. The problem is that with many nations working on AI & with many developing military systems it is now potentially an unstoppable arms race, all the way to Skynet.

Regards,

Snorvey
Lemon Quarter
Posts: 2518
Joined: November 4th, 2016, 12:51 pm
Has thanked: 827 times
Been thanked: 1008 times

Re: AI endeavours

#102682

Postby Snorvey » December 8th, 2017, 9:46 am

If, so far a long if, AI becomes self aware what would it do?

Maybe it already has and we are all living in a dream world as our comatosed bodies lie in liquid filled pods, being fed the remains of the dead whilst our bodies generate energy for the machines

dspp
Lemon Quarter
Posts: 3627
Joined: November 4th, 2016, 10:53 am
Has thanked: 2654 times
Been thanked: 856 times

Re: AI endeavours

#102733

Postby dspp » December 8th, 2017, 12:11 pm

FredBloggs wrote:How long before a machine passes the Turing test then?


Depends if it is administered by a Brexiteer or a Remainer :)

johnhemming
Lemon Quarter
Posts: 1926
Joined: November 8th, 2016, 7:13 pm
Has thanked: 4 times
Been thanked: 173 times

Re: AI endeavours

#102774

Postby johnhemming » December 8th, 2017, 3:01 pm

odysseus2000 wrote:If, so far a long if, AI becomes self aware what would it do?

An interesting question is, of course, what self-awareness means. I am sure that almost all animals have some awareness of self (or the hive).

The practical usage of AI in things like Alexa and playing various board games is the sort of thing that with not much more development can impact on the economy to a greater extent than at the moment.

odysseus2000
Lemon Quarter
Posts: 1540
Joined: November 8th, 2016, 11:33 pm
Has thanked: 364 times
Been thanked: 190 times

Re: AI endeavours

#102820

Postby odysseus2000 » December 8th, 2017, 6:25 pm

johnhemming

An interesting question is, of course, what self-awareness means. I am sure that almost all animals have some awareness of self (or the hive).


Yes, but they can't do much about their situation in life.

An AI that becomes self aware might be able to do something about anything it considers a threat.

Everything that has threatened humans has been either removed or is monitored to allow action if it becomes a threat.

Would AI consider humans like we consider vermin?

Regards,


Return to “Macro and Global Topics”

Who is online

Users browsing this forum: No registered users and 2 guests