Another thing about Musk that bothers me is his Neuralink company. He explains what it is about in his interview with Joe Rogan that was posted in this thread a few months back. He says that its reason for being is to make humans competitive with the artificial intelligence which will emerge in coming years. His plan is to develop human-computer neural interfaces which will supposedly make humans the cognitive equals of the machines.
He justifies this in argument by saying essentially that the reason that humans will be left behind in their natural state is that we cannot type fast enough/navigate browser windows fast enough or perform similar tasks fast enough. To put it into computational language, humans lack the input-output (I/O) bandwidth to compete with machines. Musk even says "it's a bandwidth issue". However, there must be, among the probably very smart employees of Neuralink, those who know that this is argument is false.
Its falsity is proved simply by considering that the human retina and optic nerve have a fairly large input bandwidth. Perhaps a megabyte per second of data through each optic nerve (and you have two). According to a Google search, the average human is only able to read about 225 words per minute (in English). There are apparently 4.79 letters in the average word in the modern written English lexicon, so this is just over a thousand letters per minute or roughly twenty characters per second. In ASCII encoding, which covers all characters and symbols used in standard English writing, each character is a single byte of data. So the average human is able to consume only about 20B/s of text, or 160 baud, leaving significant headroom in the optic nerve's throughput.
One of the earliest microprocessors (which is a computer processor integrated on a single silicon die) was the Zilog Z80, which came out 42 years ago, an 8-bit processor that was used in millions if not billions of devices. A limitation at that time was the speed of memory chips (i.e. memory bandwidth). You would normally expect an 8-bit processor to have an 8-bit arithmetic-logic unit but the designers realized that, since memory speed was limited, they could afford to run data twice through a 4-bit ALU (a technique known as bit-slicing), and thus save on the number of transistors on the processor die, increasing the number of processors they could build on a single wafer and reducing the cost per unit.
Memory is a lot faster these days, but if you increased the memory speed available to the Z80, it still could not capitalize on it because it is limited by its architecture and the technology with which it was built. You would need to overclock it, or redesign it altogether and build it with modern technology. It is similar with the human brain and increasing its own I/O bandwidth. Anybody who has driven a car on an unfamiliar route in a very busy area with lots of road signs and markings knows that their sensory input is already enough to give them "information overload".
The problem is we just do not process information fast enough, no matter how fast we can pipe it into or out of our skulls. Any attempts to improve human I/O bandwidth are going to have underwhelming end results, no matter if they are successful or not, if the goal is to make humans compete with future AI. The human brain's own short-term memory capacity, or perhaps corruption rate, seems like another obvious cognitive limitation to me. So it would appear that Musk is commanding a company filled with very smart people and has them engaged on a fool's errand, and they likely all know it.