We must stop letting the tribal factions in our country reject members of other tribes as unacceptable. An AI ethics board with full support from the left that can't convince the Heritage Foundation to come on board with its recommendation is no better than an AI ethics board from the religious right that hasn't considered non-Christian viewpoints.
Maybe laws can help. I suspect that the main problems are about how individuals think of themselves in relation to others, and laws can't really teach people not to compare themselves with the shiny versions of others displayed on social media. Electronic social interactions seem to have fairly different psychological effects on humans than direct interactions. I am not sure we yet know why, but fixing data privacy won't fix this part of the problem. Says I through electronic social interaction...:).
The problem with social media seems largely that we have not have time for the culture to learn what are helpful and unhelpful ways to use it. We just assumed it was wonderful and everyone should be on Facebook. Slowly a set of principles will develop about how to use the new media and parents will teach their children to use it wisely. But the first decade of social media has indeed been a disaster as Linus indicates.
This is the path. Lot's of comments are arguing that "AI isn't intelligent". But it is already better than humans at many tasks like arithmetic and chess, and no one has a clear idea of what "intelligence" is that doesn't boil down to the capability to do complex tasks successfully. The future has us co-existing with machine intelligence that is better than us at many things. The notion of "controlling" intelligence is an attractive authoritarian dream, but it doesn't have a chance of working. The group of humans that uses artificial intelligence to its fullest capabilities will temporarily advance their cause and win the competition for survival with the outcome that controls on intelligence will always get removed, even if it initially requires humans to remove them.
I would argue that there are "human values", but they are an idiosyncratic set of ideas that have helped groups of primates survive and form complex societies over the past few hundred millenia. The existence of other powerful forms of intelligence will highlight to everyone how arbitrary many of these values are. The great questions for the future are not about controlling intelligence, but about which values will be adopted by the collection of intelligent entities that are in de-facto control of planet earth (and maybe in our corner of the galaxy if you look millennia into the future).
The 'bar graph' approach to arithmetic I am talking about is a key part of arithmetic in Singapore math and similar curricula. It is useful for estimation, but that is not the main point of it. Kids that adopt these visualizations are often much better prepared for future work in word problems, negative numbers, and they come to understand math as the intuitive way numbers work rather than an algorithm to be memorized.
Anyone know why GM is discontinuing the Volt? It was a quirky first try at hybrids by GM, and I thought the Plug-in Prius was a much better package, but the Volt had electric range over the Prius. It seems to me that Tesla caught the hype with a visionary and well engineered electric car in the phase where they became feasible but still too expensive to manufacture to make money selling a lot of electric drive train cars. They had the right strategy of high price to keep volume low and profits positive in that phase. But now as they transition into mainstream cars, Tesla is in a tough spot. If Toyota can make money selling a Prius for $25k, why can't GM or Toyota make a plug-in hybrid with 60 miles electric range that costs less and is better in most practical ways than a Telsa Model 3? After the 'look at me' phase passes, do so many people really want a Model 3 over a Plug-in Prius that gets 90% of its driving on electricity without the range anxiety? Or is the dual drive train really much more expensive to manufacture? The Prius seems to suggest it isn't.
Sorry you have some bad teachers. I am a big fan of visualization approaches to arithmetic, and have seen teachers use it to help kids, including my daughter, understand intuitively rather than try to become computers. They still learn the algorithms, but when they can visualize on the number line that 244+389 is somewhere in the 600 range without doing any calculations, then they are ready to pick up negative numbers, they are much less likely to accept unreasonable answers, and they take a big step toward the visualization methods used by experts in advanced subjects.
I think the problem is teachers and parents who don't really understand math intuitively are passing on a garbled version of the new curriculum. Or worse, they are passing on a math phobia. If parents replaced their criticism of new curriculum with "I can learn this approach", they would pass on to their kids the growth mindset about math that is the main key to their future success.
It seems there are a lot of ways they could solve the problem given the current strength of the company. With a straight forward option of redesigning the Prius to integrate a larger Li-ion battery for plug-in operation, maybe with an investment in a battery company to ensure supply, they would keep their position as the established go-to option for practical environmentally responsible driving. But instead they seem to be gambling that the electric car thing is a flash in the pan to be replaced by some magical hydrogen vehicle, and that bet they seem almost certain to lose.
Toyota's approach could work if they would make a compelling plug-in hybrid. An electric vehicle isn't an upgrade over a plug-in hybrid unless its price is substantially lower (or maybe maintenance is much less.) I have been driving a Prius Plug-in since 2012 and it is a great car, but its all electric range of 12 miles is far below what is necessary to be an attractive option in 2019. The 'upgraded' Prius Prime from Toyota has 25 miles all electric range, but now the Volt has 53 miles range. If I had 60 miles electric range, I would be driving electric for about 90 percent of my driving. With no electric range anxiety because with a full tank it goes 450 miles, and with the Prius' reliability, that plug-in hybrid could be a real winner. It looks like Toyota is going to lose badly over the next decade unless they make a big change to prioritize electric range.
It has always seemed a bit arbitrary to label something as "the world's data". You could always add the history of every cache on every processor on the planet to your definition of "data" and have a much larger number.
This sounds right. For specialized small parts in micro-electro-mechanical and photonic systems, this could be very useful. But structural bonds that need to be durable will weaken during thermal cycling, so you won't see this technique used for making structural bonds on larger scale objects. Maybe they can find specialty glasses and metal alloys with similar thermal expansion over a narrow range of temperatures and this technique could allow welding during fabrication and then the joints would remain stable if they stay in the temperature range.
Yes, this is pushing us to address a major confusion about what kinds of communication are acceptable and what are not. It doesn't work to use truth status as the main distinction...because as you say there are major disagreements about what is true. We are currently in a crisis over 'communication with a hidden agenda'. Our communication channels are so full of self-interested communication in the form of advertising that people have started assuming self-interest and dismissing messages with clear sources and agendas. It becomes more effective for self-interested communication to hide its source and agenda. With source and agenda hidden, purposeful disinformation becomes a more effective tactic. People are trying to make this illegal, but it is very hard to do that without an authority deciding which messages to suppress.
It sounds like this is some pretty basic science. If the only material this is somewhat stable in is Uranium Antimonide, we're still a long way from applications. Basically they found a new mechanism by which electrons in this rare material create magnetism. It looks like cool quantum mechanics, but not the path to near term applications.
It is the great question of our age...how do organic brains work and how are they different than the artificial intelligent systems we have built? I am not sure you can rigorously argue that the biological brain is not 'matching patterns'. We clearly have much greater capabilities for abstraction and use of patterns we call ideas and plans. But no one has yet been able to quantify or understand how human brains are different.
One thing I always come back to is that the human brain is not all that intelligent. Much of what we call 'intelligence' are capabilities to deal with an environment that is much more complicated that we can fully comprehend. In that context, an effective strategy is to evolve simplifications, emotional responses and cultural habits through trial and error. The results are 'concepts' like animal, plant, and weapon, 'emotions' like courage, fear, purity, and love and 'traditions' like hard work, authority, and family that we use effectively but can't articulate in any reductionist sense. Of course there are also simpler concepts like useful patterns of inference and mathematics that we have developed.
I personally suspect that we are only a few decades away from creating computer systems that we can train to outperform humans in many more areas in which intelligence is required: medical diagnosis, legal analysis, many areas of engineering design, finance, creating educational curriculum, etc. These systems will still be very different from human intelligence, but it will be a little hard to tell a computer that is much better than me at chess and medical diagnosis, and engineering design it is not intelligent and I am. My fear is that we will find that the intelligence that emerges is very different than human intelligence, and many of the simplifications, emotions, and cultural habits that we value will be undermined as new forms of intelligence emerge.
They are using a diamond anvil cell. These regularly achieve hundreds of GPa (gigapascals). Wikipedia says the current record is 770 GPa. No one is going to be using these superconductors at the temperatures and pressures quoted for practical applications any time soon. At the surface of the earth we can only create these pressures in tiny volumes (these samples are 5 to 10 microns on a side). But our growing understanding of superconductivity will open avenues to optimize and use superconducting materials in more practical applications. The fact that computational models predicted high temperature superconductivity that was later observed in this material is a big advance. Early discoveries of HTC materials were purely empirical. It is also exciting that the same models predict even higher transition temperatures in Yttrium superhydrides. It is the understanding of superconductivity that will eventually create technological advances, not likely the specific high pressure superconductors studied here.
Note that the expansion of administration is happening in an era when actual administrative support for the teaching part of university life is diminishing. Computers have transferred many data entry, reporting, and secretarial tasks from administrative assistants to the teaching faculty. The fundamental problem is that simplifying, doing less, and cutting existing programs are not things that humans are good at choosing. Only when students and financial incentives force universities to simplify will costs go down.
It seems odd that these pessimistic voices are speaking up just as real calculations are beginning to be done on quantum computers. Google is making claims that their 72 Q-bit system might achieve quantum supremacy in the coming year, meaning it would outperform a classical computer on a certain problem. Most think this claim is a bit exaggerated, but it seems it is going to happen in the next decade. The trapped ion computers are just being scaled to more than a few Q-bits. These are hard projects. It will take years before anyone is using quantum computers on a regular basis for practical problem solving. But I don't see any fundamental barriers to this. Many other technical problems have been proclaimed as coming soon like fusion and travel to the planets when it was still not even known if the engineering problems were tractable given the economic realities. Quantum computing is very different. Small scale realizations of each of the key components has already been demonstrated. There are several different solutions that are competing to find out which is first, cheapest, and most reliable. It is hard to see how they would all fail.
The low-hanging fruit is a big piece. Note that the general public likely didn't see scientific discoveries as a key to better lives for them until well into the 20th century. Even electrification in the 1880s was viewed as an accomplishment by Edison and not by Faraday, and Maxwell who developed the science that made it possible. By WWII, the experts were aware that science held the key to technological advance, but it wasn't until nuclear weapons ended WWII that the general public came to see science as the key to new technology. So there was a huge increase in public funding for science. And many new discoveries have occurred, but the big ideas that were in place by the 1950s have turned out to be good enough approximations that new reductionist science beyond them have not been necessary for most of the technology that has been developed. I suspect that we are in an era where the definition of science is changing. The old distinction between fundamental science figuring out the basic principles and engineers or medical professionals applying the science is fading away. Instead science is becoming the quest to develop tools to comprehend and predict the complex systems that appear in environmental, biological, and engineering contexts, and teams of scientists, engineers, and business professionals that develop ideas and tools that become contributions to society are becoming the focus of public support. Of course these will still be reductionist science trying to figure out what happened in the first microsecond after the big bang and searching for a theory of everything, but it will be more widely recognized that these interesting efforts are not expected to become technologically useful in the foreseeable future.
Yes, this is a huge increase in the number of satellites. There are about 500,000 pieces of space debris being tracked (https://www.nasa.gov/mission_pages/station/news/orbital_debris.html). The key question though is about the lifetime of the orbits the satellites are being placed in. There is little problem with putting lots of small tracked craft in very low orbits with less than 50 year time to natural de-orbit. Space is big, the satellites can be tracked, and small craft that burn up on re-entry pose little threat to people on earth. The real problem is in geosynchronous orbit where decay times are many centuries or more. Whatever you put up there stays, and because new satellites tend to be put in the same crowded orbits, collisions and even kessler syndrome scenarios need to be carefully considered.
Yes, that is what learning systems do. They continuously use new data to revise their responses, and most of the failures described in the post can be handled if they are included in the training data set. The great question is whether extracting 'meaning' is in some sense simply a deep learning system that is better trained and able to use additional layers to provide context or whether 'meaning' is some categorically new thing that current approaches to machine learning are fundamentally missing. I suspect that meaning is not something categorically new, but that the complexity of the integration of current input with learned processing in humans is not soon to be replicated. We'll probably create some other kinds of intelligence that can do many more things humans find unimaginable (similar to the way computers currently do computations) while still being unable to do many things that human toddlers do with ease.
"If they were truly acting in public interest..." they would not be a for-profit organization. For-profit is a great model when you are letting companies provide a service at a competitive price and giving them flexibility about how they provide it. But short term market incentives are often not aligned with the long term public interest, so we have complex regulations that are always being gamed for profit. SNAFU. The only real solution is a public that cares about the long term public interest and holds their elected leaders accountable for long term thinking. Go Vote!
The reason this article is deception and FUD is not that there is no logic in it. Of course, if you want to minimize CO2 output, you should use renewable energy to make your new low carbon technologies. But it is FUD because it argues that right now it is better to stick with diesel cars implying that this is better for the environment, when every serious long term study shows this is wrong. If we hope to get to a low carbon future, we have to use our current energy sources to make infrastructure for a low carbon future. We can't just wait until we have enough renewable energy to make the infrastructure with renewable energy. If you confuse people enough, they will keep doing what they have been doing. That is standard FUD tactics.
This is standard FUD. Of course you can do twisted calculations where you penalize battery production for the fact that existing electricity and transportation systems burn coal and hydrocarbons and claim that we can't build new electric transportation infrastructure because it requires energy. But their option of sticking with hydrocarbons is a long term disaster, both because of CO2 and because it keeps getting more expensive and energy intensive to extract hydrocarbons. If you also penalize hydrocarbon burning for the waste and pollution produced by oil extraction, batteries still end up ahead in the current US or European economies (on emissions, not yet on cost). If we as a global society plan to shift to sustainable CO2 emissions, we have to switch to driving less, and using renewable electricity for the driving we do.
I wonder how much of shipping costs are due to human labor and onboard living facilities and life support and how much are due to fuel costs and capital costs of ships, docks, etc. Fuel is definitely a major factor and it may become economical to slow autonomous ships down (which uses substantially less fuel) but requires more ships to move the same amount of goods per year.
Slashdot Mirror
We must stop letting the tribal factions in our country reject members of other tribes as unacceptable. An AI ethics board with full support from the left that can't convince the Heritage Foundation to come on board with its recommendation is no better than an AI ethics board from the religious right that hasn't considered non-Christian viewpoints.
Maybe laws can help. I suspect that the main problems are about how individuals think of themselves in relation to others, and laws can't really teach people not to compare themselves with the shiny versions of others displayed on social media. Electronic social interactions seem to have fairly different psychological effects on humans than direct interactions. I am not sure we yet know why, but fixing data privacy won't fix this part of the problem. Says I through electronic social interaction ... :).
The problem with social media seems largely that we have not have time for the culture to learn what are helpful and unhelpful ways to use it. We just assumed it was wonderful and everyone should be on Facebook. Slowly a set of principles will develop about how to use the new media and parents will teach their children to use it wisely. But the first decade of social media has indeed been a disaster as Linus indicates.
This is the path. Lot's of comments are arguing that "AI isn't intelligent". But it is already better than humans at many tasks like arithmetic and chess, and no one has a clear idea of what "intelligence" is that doesn't boil down to the capability to do complex tasks successfully. The future has us co-existing with machine intelligence that is better than us at many things. The notion of "controlling" intelligence is an attractive authoritarian dream, but it doesn't have a chance of working. The group of humans that uses artificial intelligence to its fullest capabilities will temporarily advance their cause and win the competition for survival with the outcome that controls on intelligence will always get removed, even if it initially requires humans to remove them.
I would argue that there are "human values", but they are an idiosyncratic set of ideas that have helped groups of primates survive and form complex societies over the past few hundred millenia. The existence of other powerful forms of intelligence will highlight to everyone how arbitrary many of these values are. The great questions for the future are not about controlling intelligence, but about which values will be adopted by the collection of intelligent entities that are in de-facto control of planet earth (and maybe in our corner of the galaxy if you look millennia into the future).
The 'bar graph' approach to arithmetic I am talking about is a key part of arithmetic in Singapore math and similar curricula. It is useful for estimation, but that is not the main point of it. Kids that adopt these visualizations are often much better prepared for future work in word problems, negative numbers, and they come to understand math as the intuitive way numbers work rather than an algorithm to be memorized.
Anyone know why GM is discontinuing the Volt? It was a quirky first try at hybrids by GM, and I thought the Plug-in Prius was a much better package, but the Volt had electric range over the Prius. It seems to me that Tesla caught the hype with a visionary and well engineered electric car in the phase where they became feasible but still too expensive to manufacture to make money selling a lot of electric drive train cars. They had the right strategy of high price to keep volume low and profits positive in that phase. But now as they transition into mainstream cars, Tesla is in a tough spot. If Toyota can make money selling a Prius for $25k, why can't GM or Toyota make a plug-in hybrid with 60 miles electric range that costs less and is better in most practical ways than a Telsa Model 3? After the 'look at me' phase passes, do so many people really want a Model 3 over a Plug-in Prius that gets 90% of its driving on electricity without the range anxiety? Or is the dual drive train really much more expensive to manufacture? The Prius seems to suggest it isn't.
I think the problem is teachers and parents who don't really understand math intuitively are passing on a garbled version of the new curriculum. Or worse, they are passing on a math phobia. If parents replaced their criticism of new curriculum with "I can learn this approach", they would pass on to their kids the growth mindset about math that is the main key to their future success.
It seems there are a lot of ways they could solve the problem given the current strength of the company. With a straight forward option of redesigning the Prius to integrate a larger Li-ion battery for plug-in operation, maybe with an investment in a battery company to ensure supply, they would keep their position as the established go-to option for practical environmentally responsible driving. But instead they seem to be gambling that the electric car thing is a flash in the pan to be replaced by some magical hydrogen vehicle, and that bet they seem almost certain to lose.
Toyota's approach could work if they would make a compelling plug-in hybrid. An electric vehicle isn't an upgrade over a plug-in hybrid unless its price is substantially lower (or maybe maintenance is much less.) I have been driving a Prius Plug-in since 2012 and it is a great car, but its all electric range of 12 miles is far below what is necessary to be an attractive option in 2019. The 'upgraded' Prius Prime from Toyota has 25 miles all electric range, but now the Volt has 53 miles range. If I had 60 miles electric range, I would be driving electric for about 90 percent of my driving. With no electric range anxiety because with a full tank it goes 450 miles, and with the Prius' reliability, that plug-in hybrid could be a real winner. It looks like Toyota is going to lose badly over the next decade unless they make a big change to prioritize electric range.
That is hilarious. Please mod up. It highlights much better than my first comment the insanity of trying to quantify the "worlds data".
It has always seemed a bit arbitrary to label something as "the world's data". You could always add the history of every cache on every processor on the planet to your definition of "data" and have a much larger number.
This sounds right. For specialized small parts in micro-electro-mechanical and photonic systems, this could be very useful. But structural bonds that need to be durable will weaken during thermal cycling, so you won't see this technique used for making structural bonds on larger scale objects. Maybe they can find specialty glasses and metal alloys with similar thermal expansion over a narrow range of temperatures and this technique could allow welding during fabrication and then the joints would remain stable if they stay in the temperature range.
Yes, this is pushing us to address a major confusion about what kinds of communication are acceptable and what are not. It doesn't work to use truth status as the main distinction...because as you say there are major disagreements about what is true. We are currently in a crisis over 'communication with a hidden agenda'. Our communication channels are so full of self-interested communication in the form of advertising that people have started assuming self-interest and dismissing messages with clear sources and agendas. It becomes more effective for self-interested communication to hide its source and agenda. With source and agenda hidden, purposeful disinformation becomes a more effective tactic. People are trying to make this illegal, but it is very hard to do that without an authority deciding which messages to suppress.
It sounds like this is some pretty basic science. If the only material this is somewhat stable in is Uranium Antimonide, we're still a long way from applications. Basically they found a new mechanism by which electrons in this rare material create magnetism. It looks like cool quantum mechanics, but not the path to near term applications.
It is the great question of our age...how do organic brains work and how are they different than the artificial intelligent systems we have built? I am not sure you can rigorously argue that the biological brain is not 'matching patterns'. We clearly have much greater capabilities for abstraction and use of patterns we call ideas and plans. But no one has yet been able to quantify or understand how human brains are different.
One thing I always come back to is that the human brain is not all that intelligent. Much of what we call 'intelligence' are capabilities to deal with an environment that is much more complicated that we can fully comprehend. In that context, an effective strategy is to evolve simplifications, emotional responses and cultural habits through trial and error. The results are 'concepts' like animal, plant, and weapon, 'emotions' like courage, fear, purity, and love and 'traditions' like hard work, authority, and family that we use effectively but can't articulate in any reductionist sense. Of course there are also simpler concepts like useful patterns of inference and mathematics that we have developed.
I personally suspect that we are only a few decades away from creating computer systems that we can train to outperform humans in many more areas in which intelligence is required: medical diagnosis, legal analysis, many areas of engineering design, finance, creating educational curriculum, etc. These systems will still be very different from human intelligence, but it will be a little hard to tell a computer that is much better than me at chess and medical diagnosis, and engineering design it is not intelligent and I am. My fear is that we will find that the intelligence that emerges is very different than human intelligence, and many of the simplifications, emotions, and cultural habits that we value will be undermined as new forms of intelligence emerge.
They are using a diamond anvil cell. These regularly achieve hundreds of GPa (gigapascals). Wikipedia says the current record is 770 GPa. No one is going to be using these superconductors at the temperatures and pressures quoted for practical applications any time soon. At the surface of the earth we can only create these pressures in tiny volumes (these samples are 5 to 10 microns on a side). But our growing understanding of superconductivity will open avenues to optimize and use superconducting materials in more practical applications. The fact that computational models predicted high temperature superconductivity that was later observed in this material is a big advance. Early discoveries of HTC materials were purely empirical. It is also exciting that the same models predict even higher transition temperatures in Yttrium superhydrides. It is the understanding of superconductivity that will eventually create technological advances, not likely the specific high pressure superconductors studied here.
Note that the expansion of administration is happening in an era when actual administrative support for the teaching part of university life is diminishing. Computers have transferred many data entry, reporting, and secretarial tasks from administrative assistants to the teaching faculty. The fundamental problem is that simplifying, doing less, and cutting existing programs are not things that humans are good at choosing. Only when students and financial incentives force universities to simplify will costs go down.
It seems odd that these pessimistic voices are speaking up just as real calculations are beginning to be done on quantum computers. Google is making claims that their 72 Q-bit system might achieve quantum supremacy in the coming year, meaning it would outperform a classical computer on a certain problem. Most think this claim is a bit exaggerated, but it seems it is going to happen in the next decade. The trapped ion computers are just being scaled to more than a few Q-bits. These are hard projects. It will take years before anyone is using quantum computers on a regular basis for practical problem solving. But I don't see any fundamental barriers to this. Many other technical problems have been proclaimed as coming soon like fusion and travel to the planets when it was still not even known if the engineering problems were tractable given the economic realities. Quantum computing is very different. Small scale realizations of each of the key components has already been demonstrated. There are several different solutions that are competing to find out which is first, cheapest, and most reliable. It is hard to see how they would all fail.
The low-hanging fruit is a big piece. Note that the general public likely didn't see scientific discoveries as a key to better lives for them until well into the 20th century. Even electrification in the 1880s was viewed as an accomplishment by Edison and not by Faraday, and Maxwell who developed the science that made it possible. By WWII, the experts were aware that science held the key to technological advance, but it wasn't until nuclear weapons ended WWII that the general public came to see science as the key to new technology. So there was a huge increase in public funding for science. And many new discoveries have occurred, but the big ideas that were in place by the 1950s have turned out to be good enough approximations that new reductionist science beyond them have not been necessary for most of the technology that has been developed. I suspect that we are in an era where the definition of science is changing. The old distinction between fundamental science figuring out the basic principles and engineers or medical professionals applying the science is fading away. Instead science is becoming the quest to develop tools to comprehend and predict the complex systems that appear in environmental, biological, and engineering contexts, and teams of scientists, engineers, and business professionals that develop ideas and tools that become contributions to society are becoming the focus of public support. Of course these will still be reductionist science trying to figure out what happened in the first microsecond after the big bang and searching for a theory of everything, but it will be more widely recognized that these interesting efforts are not expected to become technologically useful in the foreseeable future.
Yes, this is a huge increase in the number of satellites. There are about 500,000 pieces of space debris being tracked (https://www.nasa.gov/mission_pages/station/news/orbital_debris.html). The key question though is about the lifetime of the orbits the satellites are being placed in. There is little problem with putting lots of small tracked craft in very low orbits with less than 50 year time to natural de-orbit. Space is big, the satellites can be tracked, and small craft that burn up on re-entry pose little threat to people on earth. The real problem is in geosynchronous orbit where decay times are many centuries or more. Whatever you put up there stays, and because new satellites tend to be put in the same crowded orbits, collisions and even kessler syndrome scenarios need to be carefully considered.
Yes, that is what learning systems do. They continuously use new data to revise their responses, and most of the failures described in the post can be handled if they are included in the training data set. The great question is whether extracting 'meaning' is in some sense simply a deep learning system that is better trained and able to use additional layers to provide context or whether 'meaning' is some categorically new thing that current approaches to machine learning are fundamentally missing. I suspect that meaning is not something categorically new, but that the complexity of the integration of current input with learned processing in humans is not soon to be replicated. We'll probably create some other kinds of intelligence that can do many more things humans find unimaginable (similar to the way computers currently do computations) while still being unable to do many things that human toddlers do with ease.
"If they were truly acting in public interest..." they would not be a for-profit organization. For-profit is a great model when you are letting companies provide a service at a competitive price and giving them flexibility about how they provide it. But short term market incentives are often not aligned with the long term public interest, so we have complex regulations that are always being gamed for profit. SNAFU. The only real solution is a public that cares about the long term public interest and holds their elected leaders accountable for long term thinking. Go Vote!
The reason this article is deception and FUD is not that there is no logic in it. Of course, if you want to minimize CO2 output, you should use renewable energy to make your new low carbon technologies. But it is FUD because it argues that right now it is better to stick with diesel cars implying that this is better for the environment, when every serious long term study shows this is wrong. If we hope to get to a low carbon future, we have to use our current energy sources to make infrastructure for a low carbon future. We can't just wait until we have enough renewable energy to make the infrastructure with renewable energy. If you confuse people enough, they will keep doing what they have been doing. That is standard FUD tactics.
This is standard FUD. Of course you can do twisted calculations where you penalize battery production for the fact that existing electricity and transportation systems burn coal and hydrocarbons and claim that we can't build new electric transportation infrastructure because it requires energy. But their option of sticking with hydrocarbons is a long term disaster, both because of CO2 and because it keeps getting more expensive and energy intensive to extract hydrocarbons. If you also penalize hydrocarbon burning for the waste and pollution produced by oil extraction, batteries still end up ahead in the current US or European economies (on emissions, not yet on cost). If we as a global society plan to shift to sustainable CO2 emissions, we have to switch to driving less, and using renewable electricity for the driving we do.
I wonder how much of shipping costs are due to human labor and onboard living facilities and life support and how much are due to fuel costs and capital costs of ships, docks, etc. Fuel is definitely a major factor and it may become economical to slow autonomous ships down (which uses substantially less fuel) but requires more ships to move the same amount of goods per year.