Using the Thorium fuel cycle in molten salt reactors (LFTR) has always been the way to go with nuclear power.
Indeed. Other than NOT WORKING, LFTRs are great. LFTRs are an armchair engineer's wet dream: they work fantastic in theory, as long as you ignore messy details like reality.
We can however sequester this CO2 emissions by growing plants and burying the cellulose underground, which after a few million years of lagering will become hydrocarbons.
That is silly, wasteful, and pointless. We can just sequester the CO2 directly. A shale formation that held billions of cubic meters of CH4 for millions of years will have no problem holding CO2.
There is already serious talk about electric airplanes.
Electric planes may work for short hops, like Boston to NYC, or London to Paris. But there is no way they can go from SFO to Shanghai without some profound breakthroughs. Long haul is where most aviation energy consumption occurs.
Bullshit. We can power all those items with nuclear power.
Nukes can work for cement, which just needs heat for the kiln. But nuclear aircraft? I don't think so. An iron blast furnace uses metallurgical coal (converted to coke), as an integral part of the process. You can't just drop in nuclear as a replacement.
Like TFA says, we need new tech. Business leaders and politicians can't save the world. Only nerds can do that.
That is a different project: A sensor that detects if a dumpster is full. Interesting, and perhaps useful, but not at all the same thing as a trash sorter.
They don't have high-quality service now, so I don't think they really care. I've ended up randomly with probably a hundred bucks of random stuff at my house that was misdelivered to me instead of to people with a similar house number half a mile away.
They scan the barcode at the point of delivery and use GPS to confirm the address is correct. I haven't has a misdelivery in years.
the second is hard enough that humans can't even figure it out some times, much less train a machine learning program to a high degree of accuracy.
The bot doesn't need a high degree of accuracy. If it can pull out, say, 80% of the recyclables, that will be a "good enough" solution, and a big improvement over the status quo.
All the sorter would have to do is look for and classify by for the plastic type indicator stamped into the bottom of each container.
That would only work for properly oriented bottles/cans that are clean. It would be better to focus on recognizing "bottles", "cans", bags", and other recyclables, and then look at the type indicator after it is picked up and rotated.
So why not use AI on this? I might earn 50k, but my ideas are great!
I agree that this is a good idea. Why isn't this happening?
Is anyone interested in starting an open source project to do this? We will need coders with expertise in CUDA/OpenCL, Tensorflow, etc. We also need to build up a database of training images. Oh, and some MechEngs to build the gripper.
A good rule of thumb for US (both federal and state) government projects it 3x the initial projected cost.
The problem is that the JWST has gone WAY over the normal 3x threshold. Most internal budgeting goes by the 3x rule, so that is expected and planned for, since everybody knows how the "lowball-approve-jackup" game is played. But by grossly exceeding that threshold, JWST has become known as "the telescope that ate astronomy" since so much money has been diverted from other projects.
Indeed. The security ramifications were immediately pointed out by many people as soon as this idiotic proposal was made. But it went forward anyway so they could sell new domain names, and force legitimate companies to spend even more to buy up every possible permutation of their names.
The only good solution now is for browsers to block these domains, or at least throw up a flashing SCAM warning whenever one is accessed.
the employer has probably not paid for training or made any other investment in me.
Years ago it was common for employers to invest in training because they could require the employee to agree to either continue to work for X years, or reimburse the company for the cost of the training if they quit early.
Today, those agreements are illegal. So why should a company invest in you if they don't know if they can recoup the cost?
The change in the law was to "protect employees". But the result was lower skills, lower productivity, lower pay, hurting employees, hurting companies, and hurting the country. Oops. Another example of unintended consequences.
deadwood will continue to be paid what they're worth (or not) as well.
I don't think so. Employers will start gaming this as soon as it goes live, by writing glowing endorsements for their deadwood employees in the hope that someone else poaches them. That way they can get rid of them without paying severance.
I'm still not sure that it's true "intelligence" though...
Of course not, except in a very narrow sense. AI is a field of research to develop machine intelligence. We are making progress, but it will be a long journey.
But what *is* intelligence, exactly?
Intelligence is the ability to formulate an effective initial response to a novel situation.
The wording here is important. It is an "ability" not a mechanism. A system that consistently behaves intelligently is intelligent. The internal mechanism is irrelevant. It is an ability to "formulate" a plan, not a physical ability to act on the plan. It is the initial response that matters, so random trial and error, or undirected evolution, don't count. It is the response to "novel" situations that matter, not just the ability to lookup and repeat previous solutions.
What is "effective" may be subjective, but in this case it is obviously to win the game.
Slashdot Mirror
Using the Thorium fuel cycle in molten salt reactors (LFTR) has always been the way to go with nuclear power.
Indeed. Other than NOT WORKING, LFTRs are great. LFTRs are an armchair engineer's wet dream: they work fantastic in theory, as long as you ignore messy details like reality.
A nuclear plant could easily power a processing plant to produce methane from the CO2 in the air and water (Sabatier reaction).
Sure, if by "easily" you mean complicated and extremely expensive. Or are we assuming that nuclear power is "too cheap to meter"?
Many steel plants already use induction furnaces for for melt processes
Those are "mini-mills". They make steel from steel. They don't make steel from ore.
but the addition of coke to remove impurities is a required part of the process.
The coke removes the oxygen, which is MOST of the iron ore. It is the iron that is the impurity.
We can however sequester this CO2 emissions by growing plants and burying the cellulose underground, which after a few million years of lagering will become hydrocarbons.
That is silly, wasteful, and pointless. We can just sequester the CO2 directly. A shale formation that held billions of cubic meters of CH4 for millions of years will have no problem holding CO2.
There is already serious talk about electric airplanes.
Electric planes may work for short hops, like Boston to NYC, or London to Paris. But there is no way they can go from SFO to Shanghai without some profound breakthroughs. Long haul is where most aviation energy consumption occurs.
Bullshit. We can power all those items with nuclear power.
Nukes can work for cement, which just needs heat for the kiln. But nuclear aircraft? I don't think so. An iron blast furnace uses metallurgical coal (converted to coke), as an integral part of the process. You can't just drop in nuclear as a replacement.
Like TFA says, we need new tech. Business leaders and politicians can't save the world. Only nerds can do that.
You don't know how to use Google, do you. http://lmgtfy.com/?q=NCSU+ECE+...
It's the second link.
That is a different project: A sensor that detects if a dumpster is full. Interesting, and perhaps useful, but not at all the same thing as a trash sorter.
They don't have high-quality service now, so I don't think they really care. I've ended up randomly with probably a hundred bucks of random stuff at my house that was misdelivered to me instead of to people with a similar house number half a mile away.
They scan the barcode at the point of delivery and use GPS to confirm the address is correct. I haven't has a misdelivery in years.
An NCSU ECE Senior Design team last year built a recycling sorter
A Google search turns up nothing. Do you have a link?
a useful solution is a dissertation away for PhD candidate.
Hee hee. "Useful solution" and "PhD candidate" in the same sentence. That's funny.
Phoenix has had several instances over the years of murderers literally throwing away their victims, in the municipal collection.
Those could be sorted into a general "compostable" bin.
the second is hard enough that humans can't even figure it out some times, much less train a machine learning program to a high degree of accuracy.
The bot doesn't need a high degree of accuracy. If it can pull out, say, 80% of the recyclables, that will be a "good enough" solution, and a big improvement over the status quo.
All the sorter would have to do is look for and classify by for the plastic type indicator stamped into the bottom of each container.
That would only work for properly oriented bottles/cans that are clean. It would be better to focus on recognizing "bottles", "cans", bags", and other recyclables, and then look at the type indicator after it is picked up and rotated.
So why not use AI on this? I might earn 50k, but my ideas are great!
I agree that this is a good idea. Why isn't this happening?
Is anyone interested in starting an open source project to do this? We will need coders with expertise in CUDA/OpenCL, Tensorflow, etc. We also need to build up a database of training images. Oh, and some MechEngs to build the gripper.
Also, JWST is WAY more capable that Hubble. It is like comparing apples to watermelons.
A good rule of thumb for US (both federal and state) government projects it 3x the initial projected cost.
The problem is that the JWST has gone WAY over the normal 3x threshold. Most internal budgeting goes by the 3x rule, so that is expected and planned for, since everybody knows how the "lowball-approve-jackup" game is played. But by grossly exceeding that threshold, JWST has become known as "the telescope that ate astronomy" since so much money has been diverted from other projects.
Saw this coming years ago.
Indeed. The security ramifications were immediately pointed out by many people as soon as this idiotic proposal was made. But it went forward anyway so they could sell new domain names, and force legitimate companies to spend even more to buy up every possible permutation of their names.
The only good solution now is for browsers to block these domains, or at least throw up a flashing SCAM warning whenever one is accessed.
the employer has probably not paid for training or made any other investment in me.
Years ago it was common for employers to invest in training because they could require the employee to agree to either continue to work for X years, or reimburse the company for the cost of the training if they quit early.
Today, those agreements are illegal. So why should a company invest in you if they don't know if they can recoup the cost?
The change in the law was to "protect employees". But the result was lower skills, lower productivity, lower pay, hurting employees, hurting companies, and hurting the country. Oops. Another example of unintended consequences.
deadwood will continue to be paid what they're worth (or not) as well.
I don't think so. Employers will start gaming this as soon as it goes live, by writing glowing endorsements for their deadwood employees in the hope that someone else poaches them. That way they can get rid of them without paying severance.
Ironically, the biggest supporters of free trade were the Big Cotton slaveowners.
Why is that ironic? Are you implying that the slave owners were progressive liberals?
You might consider that there is no left or right.
You might consider that, as the centuries go by, parties realign and change constituencies.
Some how is this is the liberals doing?
At least in America, protectionism is popular with the left, and much less so with conservatives.
Trumpism is a blend of the stupidest policies from both left and right.
The latest distinction between machine learning and AI that I saw ...
Where did you see that? ML is a proper subset of AI. Period.
Anything that is ML is also AI. But there are subfields of AI, such as min-max and alpha-beta pruning, that are not ML.
most domestic food animals live good lives
Have you ever been in a battery cage warehouse? The noise is deafening. The stench will likely make you vomit.
The first thing I thought of was The Giving Plague by David Brin
In real life there are multiple factors that can disqualify you from donating blood. Having received a blood transfusion is one of them.
In the US, about 10% of all expenditures go to physician fees
Do you have a citation for this factoid? It seems wildly implausible.
I'm still not sure that it's true "intelligence" though...
Of course not, except in a very narrow sense. AI is a field of research to develop machine intelligence. We are making progress, but it will be a long journey.
But what *is* intelligence, exactly?
Intelligence is the ability to formulate an effective initial response to a novel situation.
The wording here is important. It is an "ability" not a mechanism. A system that consistently behaves intelligently is intelligent. The internal mechanism is irrelevant. It is an ability to "formulate" a plan, not a physical ability to act on the plan. It is the initial response that matters, so random trial and error, or undirected evolution, don't count. It is the response to "novel" situations that matter, not just the ability to lookup and repeat previous solutions.
What is "effective" may be subjective, but in this case it is obviously to win the game.