-
Posts
2,465 -
Joined
-
Last visited
-
Days Won
219
Content Type
Profiles
Forums
Blogs
Everything posted by efaardvark
-
Found out I "get" to work 2 extra days this week. Joy.
At least that means my weekends are back in sync with the rest of the world's.
-
Lol...at least it's only 2 days and not practically the whole month. Been crazy here, like some kid kicked over our anthill and we're scrambling all over the place trying to put it back up again. I'm so tired.
Enjoy the weekends are you having to meet a deadline or did something else come up at work?
-
Kind of a combination of factors here. Higher than average workload this year with chandryaan-2 launch and testing/training for them plus a bunch of cubes going up on the one hand. On the other, in the last several months we've also lost 2 workers. 3 actually, one retired, one got fired, and then one of the replacements didn't work out & we had to start over. We're a 2-position, 24/7 operation so losing two out of 8 people .. we're feeling it. 2019 has not been a good year so far.
I talked to my boss today & he says he's hired a replacement for the replacement so there's light at the end of the tunnel. However, even best case it'll take a couple weeks for badging, background check, and other bureaucrazy before he can even start training for the actual job, and another couple/few months after that before he's pulling his weight.
-
-
-
Well, looks like Chandrayaan-2 launched successfully. Launches are fun to watch, but the DSN's job doesn't really start until they actually deploy and we start getting data on the ground, and the first few tracks after launch are apt to be .. interesting until the critical spacecraft checkout is done and things settle into a routine.
At least it is a swing shift schedule for me so I got to sleep in. My boss had to be there at 2am for the launch. OTOH, I'm once again batting cleanup so I get to deal with any issues not caught in pre-launch testing that crop up.
-
-
Wow..
-
I think my two cents might actually be relevant here My comment in one sentence would be that we are not even close to being ready to actually implement something like this in any reliable way.
As it happens I both studied and now work in a closely-related field (although my focus has changed from empirical to theoretical/philosophical). I was quite enthusiastic in the beginning, but I can say with quite a high level of confidence afters spending some years in that field that we don't understand the brain. We have accumulated a lot of both empirical data and models, and we do know a lot more then just 100 or even 10 years ago. But we are far from knowing how the brain works when it comes to any function that is a bit more complex than simple neural networks or even action potentials (even those are very tricky if they deviate from the prototypical neurons usually studied).
When it comes to BMI specifically, there is some interesting stuff going on right now. One of the research groups at my university is working in a particular field regarding BMI which got me interested in it, and we even established a new research group dealing with philosophical problems related to it (although I would not say that I have much expertise when it comes to the technical aspect of BMI). What Neurolink is showing here is promising in achieving the goal of activating certain sets of neurons but their possible application (at least as presented here) a bit oversimplifying how certain functions are represented in the brain (especially when it comes to localization). The basic idea is not new, but how they are implementing ('wireless chips') is something that has been notoriously difficult (but has been researched for quite a few years).When you take these aspects together, I think even if we successfully implanted even much more complex devices, we are still far from knowing how to use them in order to achieve something we want ("to tap into those representations") in any reliable way (meaning without serious dysfunction or non-functions). This can be used for research which might one day enable a much more successful implementation of course, but we should be realistic as to the actual mid-term achievements we can expect from it, given the current state of knowledge. This is disregarding the huge ethical issues concerning malfunctions and risks.
So yeah, interesting stuff but as it often is with publications or innovations, the actual immediate usefulness is still quite 'hypothetical'.
-
I totally agree that we don't understand the brain. It has had literally hundreds of millions of years to evolve, and genetic algorithms are notorious for producing idiosyncratic designs. I'm reminded of one research project that used a large chip with an array of identical circuits that was used to develop some software using a "genetic" process to evolve the software. Each bit of software was slightly modified/mutated and assigned one of the array elements to use, then evaluated against the software design criteria. Evolve, repeat. It worked extremely well, with the best software being roughly 5x as good as the software developed using more conventional means. Then they tried to figure out why. Turns out that the "evolved" software had figured out that the supposedly identical and independent hardware cells were NOT compeletly independent and was actually using the undocumented interaction between cells to achieve its results.
So yeah, there have been and will be moments of confusion and misunderstanding on a science and engineering level. Politics and public sentiment will no doubt also play a big role beyond that. (Though both will be decades behind in actual understanding that's never really mattered before.) That said, there is progress being made as well. I view this as basic research. There are goals, and the researchers know enough to think and expect those goals are attainable. I tend to agree, but there's a HUGE grey area (no pun intended) that they need to navigate through before they get where they want to go, and plenty of opportunity for things like frustration, serendipity, and even tragedy along the way. It will not be easy or quick. It will definitely be interesting, and very likely extremely useful, though perhaps not exactly in the way(s) expected at this point.
I do think that the gross motor areas they are targeting now is a good first step. They're not trying to put anything INTO the brain, just read the motor signals that the brain is outputting. That's ambitious enough, but also limited enough, that they can make progress towards their goals in a reasonable time frame, and if they reach those then they'll be in a good position to move on from there. Clearly though Musk is thinking beyond that however. "Having the option to merge with AI is important."
-
the grey area pun was hilarious
I totally agree. And I think targeting something like the motor cortex is a pretty reasonable starting point since it at least seems to be less complex than, say, the hippocampus (or at least the parts they are investigating).
This reminds me of people who lack substantial parts of their brain but show little to no impairment. A senior neuroscientists in our department told us of a case from someone who literally lacked most of what we consider to be a full brain but showed almost no symptoms. Such cases make brain science quite difficult because of the huge flexibility it portrays. But this is not to disregard the huge advancements, both in research and treatment, made in the last decades.Merging with AI, at least what people would usually understand, is however something I don't think will happen in our lifetime. Would be extremely interesting though to see what kind of effects it would have.
-
-
A Joule is a measure of work energy. It is about equal to the kinetic energy of an apple dropped from a height of 1 meter as it hits the ground. A human being radiates about 60 joules per second of heat energy.
A Joule is equivalent to 1 watt for 1 second, or a watt-second. A watt-second uses the same units as the kilowatt-hours on your electricity bill, and the power company could just as easily bill you in Joules as in kWh. If you use 1 kWh then you've used enough energy to light a 100W lightbulb for about 10 hours, or about 3.6 million Joules, or enough to get my car about 3.3 miles down the road.
Yes, it has been a slow shift, now that you mention it.