Artificial Intelligence

[irvinehomeowner]

It?s not as easy as we hope it will be. Even Ford said they may have overestimated autonomous driving:
https://www.engadget.com/2019/04/10/ford-ceo-says-the-company-overestimated-self-driving-cars/

Kenkoko?s daughter may need to learn to drive after all.

Her mom's a Tesla die hard so probably not 

You probably should take your world trend from an leading automaker instead of a lagger. Elon Musk says Tesla will go fully autonomous by end of 2019. I hate to knock Ford but they are really the bottom fedder of the industry. There's a reason why Ford's stock is down 20% since Tesla's IPO and Tesla has gone up more than 10x

https://cleantechnica.com/2019/02/20/elon-musk-full-self-driving-teslas-this-year-unequivocal-tesla-autopilot-improves-safety/

Tesla a leading automaker over Ford? Last global count has Ford leading by quite a large margin over Tesla. And how many times has Elon missed his delivery dates?

For technology, Tesla might be more advanced than Ford, but practical realistic appolication-wise, Ford has more experience. One thing to note about autonomous driving, until it's 100% saturation, there is going to be issues because people can't deal with AI and vice versa. Have you ever been rear-ended? AI is not going to be able to stop that if the other car is driven by a person.

But, at least Ai has developed a new sport... Speedgate:
https://www.fastcompany.com/90333738/heres-how-to-play-speedgate-the-first-sport-designed-by-ai

 

[fortune11]

I wonder when elementary, middle and high schools will transition to a live video teaching model and drop the teachers and maybe just have a teachers aid in the classroom, or perhaps this is already happening? I don?t have kids that age yet,

Don?t some universities offer online degrees with live instruction? I?ve done continuing education with remote live instructors. Some dude in Indiana with a bunch of people online watching live. Then they have a queue for questions.

Although I imagine parents would struggle with this concept especially for non high school aged children. And you know the government would reduce taxes to reflect the lower payroll/benefits/pension costs.

Teacher unions will fight it!

Even if it does happen , elementary and middle schools that actually have physical instruction will become ?elite? and will have the same mad dash that we now have for college admissions

Online / remote is about cost cutting , first and foremost . All the other reasons are just PR to make it easier to force it down

 

[Irvinecommuter]

I wonder when elementary, middle and high schools will transition to a live video teaching model and drop the teachers and maybe just have a teachers aid in the classroom, or perhaps this is already happening? I don?t have kids that age yet,

Don?t some universities offer online degrees with live instruction? I?ve done continuing education with remote live instructors. Some dude in Indiana with a bunch of people online watching live. Then they have a queue for questions.

Although I imagine parents would struggle with this concept especially for non high school aged children. And you know the government would reduce taxes to reflect the lower payroll/benefits/pension costs.

It's not a good model...especial at lower grades.  Kids need that interaction and guidance. 

In Asian, video tutoring is very popular but it's I find it very dubious. 

 

[Irvinecommuter]

Many service jobs in the FIRE employment category will be "buggy whipped" by A.I. but full replacement is a tough sell. Example: If A.I. was transforming Real Estate by providing a cost crushing solution, why would Redfin see the need to partner up like this:

https://rismedia.com/2019/03/18/redfin-re-max-enter-exclusive-referral-relationship/

Boots on the ground are still needed.

P.S. Not to move the subject too far, but anyone else besides the Soylent Green family treasure the overlong, but weird and wonderful Spielberg film "A.I."?

My .02c

Hmmm, this all those buyers at the low end of the spectrum that Redfin does not/can't service because the total commission is too low to provide a commission rebate to the buyer under the Redfin minimum model.

I see RE agents similar to stock trading...industry believed that financial advisers were too valuable to replace but then Etrade, Scotttrader, etc. showed that it wasn't.

RE will actually more needed for highend of the market...middle/upper middle class will need a RE agent less and less

 

[Irvinecommuter]

It?s not as easy as we hope it will be. Even Ford said they may have overestimated autonomous driving:
https://www.engadget.com/2019/04/10/ford-ceo-says-the-company-overestimated-self-driving-cars/

Kenkoko?s daughter may need to learn to drive after all.

Her mom's a Tesla die hard so probably not 

You probably should take your world trend from an leading automaker instead of a lagger. Elon Musk says Tesla will go fully autonomous by end of 2019. I hate to knock Ford but they are really the bottom fedder of the industry. There's a reason why Ford's stock is down 20% since Tesla's IPO and Tesla has gone up more than 10x

https://cleantechnica.com/2019/02/20/elon-musk-full-self-driving-teslas-this-year-unequivocal-tesla-autopilot-improves-safety/

Tesla a leading automaker over Ford? Last global count has Ford leading by quite a large margin over Tesla. And how many times has Elon missed his delivery dates?

For technology, Tesla might be more advanced than Ford, but practical realistic appolication-wise, Ford has more experience. One thing to note about autonomous driving, until it's 100% saturation, there is going to be issues because people can't deal with AI and vice versa. Have you ever been rear-ended? AI is not going to be able to stop that if the other car is driven by a person.

But, at least Ai has developed a new sport... Speedgate:
https://www.fastcompany.com/90333738/heres-how-to-play-speedgate-the-first-sport-designed-by-ai

Considering Ford is pretty much getting out of car building industry...I wouldn't be so reliant on their views of auto AI. 

Tesla has had its problem but it fundamentally broke the model for electric cars and has made cars that auto makers claims were not possible.

 

[nosuchreality]

Given the number of Waymo cars tooling around Tempe, it's here.  Granted, they apparently killed a woman in the middle of the street last year.
https://www.nytimes.com/interactive/2018/03/20/us/self-driving-uber-pedestrian-killed.html


I suspect virtually any driver would have hit her it's just a question of how much brake they would have gotten on first.

 

[Irvinecommuter]

Given the number of Waymo cars tooling around Tempe, it's here.  Granted, they apparently killed a woman in the middle of the street last year.
https://www.nytimes.com/interactive/2018/03/20/us/self-driving-uber-pedestrian-killed.html


I suspect virtually any driver would have hit her it's just a question of how much brake they would have gotten on first.

Obviously..AI still has issues but look at development of AI from a game theory POV...people thought that AI would never beat humans in chess (which it has now done easily).  Then people said that AI could never win at Go, which it now has.  Basically...AI learns from its mistakes and everyone else's mistake.  It can take in every single game of Go played and build an algorithm to deal with every situation.  It's just a matter of time.

AI is unaffected by exterior factors like alcohol, tiredness, or loud children.  It can focus on its task with exact precision.  It's just a matter of time for AI driving.

 

[irvinehomeowner]

Given the number of Waymo cars tooling around Tempe, it's here.  Granted, they apparently killed a woman in the middle of the street last year.
https://www.nytimes.com/interactive/2018/03/20/us/self-driving-uber-pedestrian-killed.html


I suspect virtually any driver would have hit her it's just a question of how much brake they would have gotten on first.

Obviously..AI still has issues but look at development of AI from a game theory POV...people thought that AI would never beat humans in chess (which it has now done easily).  Then people said that AI could never win at Go, which it now has.  Basically...AI learns from its mistakes and everyone else's mistake.  It can take in every single game of Go played and build an algorithm to deal with every situation.  It's just a matter of time.

AI is unaffected by exterior factors like alcohol, tiredness, or loud children.  It can focus on its task with exact precision.  It's just a matter of time for AI driving.

Game theory is different from controlling over 4000 lbs of possible destruction and death. AI may be able ignore factors within its own vehicle but may not be able to deal with what happens with non-AI controlled vehicles.

Let's not overlook the fact that safety is a huge hurdle here.

I think there has to be some infrastructure changes with the roads to allow for safe AI-driven vehicles... kind of like what we saw in Minority Report (that movie had a few "futuristic" ideas that are now in use).

 

[Irvinecommuter]

Given the number of Waymo cars tooling around Tempe, it's here.  Granted, they apparently killed a woman in the middle of the street last year.
https://www.nytimes.com/interactive/2018/03/20/us/self-driving-uber-pedestrian-killed.html


I suspect virtually any driver would have hit her it's just a question of how much brake they would have gotten on first.

Obviously..AI still has issues but look at development of AI from a game theory POV...people thought that AI would never beat humans in chess (which it has now done easily).  Then people said that AI could never win at Go, which it now has.  Basically...AI learns from its mistakes and everyone else's mistake.  It can take in every single game of Go played and build an algorithm to deal with every situation.  It's just a matter of time.

AI is unaffected by exterior factors like alcohol, tiredness, or loud children.  It can focus on its task with exact precision.  It's just a matter of time for AI driving.

Game theory is different from controlling over 4000 lbs of possible destruction and death. AI may be able ignore factors within its own vehicle but may not be able to deal with what happens with non-AI controlled vehicles.

Let's not overlook the fact that safety is a huge hurdle here.

I think there has to be some infrastructure changes with the roads to allow for safe AI-driven vehicles... kind of like what we saw in Minority Report (that movie had a few "futuristic" ideas that are now in use).

It's the same.  Anticipation of future events and changes and adapting accordingly...humans are terrible at multitasking while a computer is not.  Humans are affected by things...humans are not.  All you really need to do is to upload every single accident report, investigation, causes, and outcome into an AI and it can deal with better than humans.  This is not a fight for zero accident...AI just needs to do better than humans.

 

[nosuchreality]

Given the number of Waymo cars tooling around Tempe, it's here.  Granted, they apparently killed a woman in the middle of the street last year.
https://www.nytimes.com/interactive/2018/03/20/us/self-driving-uber-pedestrian-killed.html


I suspect virtually any driver would have hit her it's just a question of how much brake they would have gotten on first.

Obviously..AI still has issues but look at development of AI from a game theory POV...people thought that AI would never beat humans in chess (which it has now done easily).  Then people said that AI could never win at Go, which it now has.  Basically...AI learns from its mistakes and everyone else's mistake.  It can take in every single game of Go played and build an algorithm to deal with every situation.  It's just a matter of time.

AI is unaffected by exterior factors like alcohol, tiredness, or loud children.  It can focus on its task with exact precision.  It's just a matter of time for AI driving.

Game theory is different from controlling over 4000 lbs of possible destruction and death. AI may be able ignore factors within its own vehicle but may not be able to deal with what happens with non-AI controlled vehicles.

Let's not overlook the fact that safety is a huge hurdle here.

I think there has to be some infrastructure changes with the roads to allow for safe AI-driven vehicles... kind of like what we saw in Minority Report (that movie had a few "futuristic" ideas that are now in use).

Safety goes to AI, already.  With the exception of the person autopiloting the Tesla in Florida, virtually all of the AI car accidents arehuman caused.  Their accident rate per mile is far lower than human driven.

The sensor didnt pick the woman in the video up. You can slow it down,frame by frame.  Time from her white shoes being visible to impact is under 1.5 seconds.

Two planes have been crashed by autopilot recently.  How manybhave crash due to pilot error?

 

[irvinehomeowner]

I think you both are missing my point.

Teaching a computer to play a game, where no lives are at risk is easily acceptable.

Teaching a computer to drive a car or pilot a plane, there are many other factors that need to be taken into account due to interaction with numerous external elements that does not occur in a game of chess or Go.

And that's why I mentioned 100% saturation or change in road infrastructure, as long as non-AI elements are involved (ie humans) there will be an issue of safety that will be difficult to get over.

I also wonder about liability. In these cases where autonomous cars are involved in accidents, who got the blame? The human occupant? The AI? The car company?

 

[inv0ke-epipen]

I think you both are missing my point.

Teaching a computer to play a game, where no lives are at risk is easily acceptable.

Teaching a computer to drive a car or pilot a plane, there are many other factors that need to be taken into account due to interaction with numerous external elements that does not occur in a game of chess or Go.

And that's why I mentioned 100% saturation or change in road infrastructure, as long as non-AI elements are involved (ie humans) there will be an issue of safety that will be difficult to get over.

I also wonder about liability. In these cases where autonomous cars are involved in accidents, who got the blame? The human occupant? The AI? The car company?

People working on self driving cars not only know about this, but have been thinking about and working on this for nearly a decade.

Google has basically solved handling other drivers, and is in the QA/proving out phase now.

Nice video diving into a few of their techniques, including simulation:

https://www.youtube.com/watch?v=Q0nGo2-y0xY&t=2544s

 

[inv0ke-epipen]

Given the number of Waymo cars tooling around Tempe, it's here.  Granted, they apparently killed a woman in the middle of the street last year.
https://www.nytimes.com/interactive/2018/03/20/us/self-driving-uber-pedestrian-killed.html


I suspect virtually any driver would have hit her it's just a question of how much brake they would have gotten on first.

Waymo != Uber. Uber's program is garbage. Waymo would not have hit her, nor would most human drivers. Uber's initial video showing the hit was super misleading as they darkened the video to make it look harder to see the pedestrian than it actually was.

Uber has since been banned from testing in the state of AZ, while Waymo continues to expand operations.

 

[irvinehomeowner]

Also, I don't think AI is not capable of what everyone is hoping.

What I am (and others) are cautious about is how to adopt AI into society safely.

And no one has addressed the issue of hacking and using AI for more nefarious purposes... HAL-9000 isn't just rainbows and unicorns.

 

[Irvinecommuter]

Also, I don't think AI is not capable of what everyone is hoping.

What I am (and others) are cautious about is how to adopt AI into society safely.

And no one has addressed the issue of hacking and using AI for more nefarious purposes... HAL-9000 isn't just rainbows and unicorns.

of course...those are all issues but industry is not going to stop in the interim. 

Again...we are not living in a world where accidents don't happen....so AI's job is not to eliminate accident...just reduce them.

 

[Irvinecommuter]

I think you both are missing my point.

Teaching a computer to play a game, where no lives are at risk is easily acceptable.

Teaching a computer to drive a car or pilot a plane, there are many other factors that need to be taken into account due to interaction with numerous external elements that does not occur in a game of chess or Go.

And that's why I mentioned 100% saturation or change in road infrastructure, as long as non-AI elements are involved (ie humans) there will be an issue of safety that will be difficult to get over.

I also wonder about liability. In these cases where autonomous cars are involved in accidents, who got the blame? The human occupant? The AI? The car company?

Teaching a computer how to drive is probably a lot easier than go or chess...think about about how many people can drive versus how many people can master chess. 

AI will be better and faster at processing data and converting into action...it never takes its "eyes" off the road and is never distracted.  It can also link up with other smart cars on the road to get a much better picture than a human can. 

 

[inv0ke-epipen]

Also, I don't think AI is not capable of what everyone is hoping.

What I am (and others) are cautious about is how to adopt AI into society safely.

And no one has addressed the issue of hacking and using AI for more nefarious purposes... HAL-9000 isn't just rainbows and unicorns.

There exists an entire industry dedicated to address the risk of hacking.

KeenLab does some great stuff targeting autopilot:https://keenlab.tencent.com/en/2019...imental-Security-Research-of-Tesla-Autopilot/

 

[irvinehomeowner]

I think you both are missing my point.

Teaching a computer to play a game, where no lives are at risk is easily acceptable.

Teaching a computer to drive a car or pilot a plane, there are many other factors that need to be taken into account due to interaction with numerous external elements that does not occur in a game of chess or Go.

And that's why I mentioned 100% saturation or change in road infrastructure, as long as non-AI elements are involved (ie humans) there will be an issue of safety that will be difficult to get over.

I also wonder about liability. In these cases where autonomous cars are involved in accidents, who got the blame? The human occupant? The AI? The car company?

Teaching a computer how to drive is probably a lot easier than go or chess...think about about how many people can drive versus how many people can master chess. 

I disagree. The reason why people can't master chess is because they don't have the memory or capacity to understand the rules and think many scenarios ahead and to be able to catalog and combine on all those scenarios into a winning outcome. Chess is like a closed box with counters to all moves because each chess piece has a limited move set and each player has to take a turn. That's what computers are very good at... just like math.

Driving however is not something you can predict because there are numerous more variables that can affect driving. Other drivers, road conditions, weather, lights, etc... it's a whole different animal.

AI will be better and faster at processing data and converting into action...it never takes its "eyes" off the road and is never distracted.  It can also link up with other smart cars on the road to get a much better picture than a human can. 

Sure, I understand that. But that speaks again to saturation and infrastructure. Cars can't link up properly if not all cars are smart or if the drivers choose not to share their information (personally, I wouldn't want another vehicle to have access to my data). It's just like those Tile tracking devices that can find your item but it relies on other people to share their Tile location services (which goes back to my whole premise of data sharing and privacy).

 

[irvinehomeowner]

Also, I don't think AI is not capable of what everyone is hoping.

What I am (and others) are cautious about is how to adopt AI into society safely.

And no one has addressed the issue of hacking and using AI for more nefarious purposes... HAL-9000 isn't just rainbows and unicorns.

There exists an entire industry dedicated to address the risk of hacking.

KeenLab does some great stuff targeting autopilot:https://keenlab.tencent.com/en/2019...imental-Security-Research-of-Tesla-Autopilot/

Exactly.

I didn't meant no one in the industry has addressed this, I mean the members here have not acknowledge that danger as credible when I brought it up here.

That article makes me trust autonomous driving less.

 

[Irvinecommuter]

I disagree. The reason why people can't master chess is because they don't have the memory or capacity to understand the rules and think many scenarios ahead and to be able to catalog and combine on all those scenarios into a winning outcome. Chess is like a closed box with counters to all moves because each chess piece has a limited move set and each player has to take a turn. That's what computers are very good at... just like math.

Driving however is not something you can predict because there are numerous more variables that can affect driving. Other drivers, road conditions, weather, lights, etc... it's a whole different animal.

Your brain operates very similar to a computer when it takes on a task like driving.  You take your past experience and combine it with your senses and combine it with reaction times.  It's why people who have never driven in the snow...don't know how to do it.  Or people who think rain is acid in Southern California. 

Computers will be much more attentive, unaffected by side issues/factors, have much better sensors, have a far bigger (and ever increasing) experience than any human, and far better reaction time than human beings. 

Chess and go are also not just about rote memory...it's about anticipation and setting up your opponent.  People thought that AI would not grasp Go because it requires a big picture view about what is happening.  AI looks silly in the beginning but it can learn nonstop and build off of real life/human experiences.
https://www.sciencedaily.com/releases/2019/01/190125094230.htm

Sure, I understand that. But that speaks again to saturation and infrastructure. Cars can't link up properly if not all cars are smart or if the drivers choose not to share their information (personally, I wouldn't want another vehicle to have access to my data). It's just like those Tile tracking devices that can find your item but it relies on other people to share their Tile location services (which goes back to my whole premise of data sharing and privacy).

As for the smart car grid, of course having a few in the early phases will not have that much of an effect but right now, we are working on zero connectiion between the cars.  Having 10 to 15% of the cars connect to each other will be an immense increase in safety and efficiency. 

It's the wave of the future...trucks already have built in GPS and black boxes to track all sorts of things.  Look at a Tesla, the app will tell you where the car is at all times and if driving, what speeds and where.  Tesla already has that information.