The Orion's Arm Universe Project Forums





<Link> Thalience and transapient perspective
#1
Thalience and transapient perspective

Science fiction author karl schroeder writes about how AIs might see the world
http://www.kschroeder.com/my-books/ventus/thalience
Reply
#2
Hmm. Interesting. Perhaps something like this (non-human minds creating different models of the universe) explains why transapient science is beyond human comprehension yet still works in our universe. Transapient tech is built on models that our minds simply aren't equipped to grasp.

Just a thought,

Todd
Reply
#3
That was an enjoyable read, nice to see the deep thoughts of intelligent people.

However, I don't completely agree with him.

For one, Quantum Mechanics and Relativity aren't two theories that explain the same thing, they're two theories that explain the universe at wildly different scales. It's not that they are irreconcilable (we don't have enough information to determine this yet), but that we haven't figured out how to join them together yet.

Loop Quantum Gravity and String Theory would be two different ways of explaining the same thing.

Supposing that there are multiple theories that explain the universe equally well, Occam's Razor comes into play. Not picking a vision of the universe that tickles your ego may be too austere for many people, but it's the rational choice. If there are multiple theories that have an equal number of assumptions, then one could pick the one that they like most, being mindful that the others have equal probability of being the truth.

Multiple working theories of the universe would not allow the creation of scientifically valid religions. Religions, by definition, assume the existence of entities, states of being, etc., that are untestable because they lie outside physical reality. Religious assumptions have no predictive ability. It would be possible to create religions that are consistent with observed reality; for example, it's possible that Buddhism is true, and people can reach Nirvana after death, but it's not possible that Fundamentalist Abrahamic Religions are true because the accounts in Genesis completely contradict the physical and secular-historical evidence. There's no hint that Nirvana actually exists, so Buddhism isn't scientifically valid.
Reply
#4
Interesting piece. I find the idea of automated science fascinating and a have often wondered if there will be a time in my life when a completely robotic team composes and tests a hypothesis. I have an amusing image of seeing a fully robot researched and written paper arrive on my desk and choosing that moment to start planning retirement :p

Incidentally the novel this concept is based on is called Ventus and is available for free on the authors website. I'm reading it now and it seems quite gripping, if a little outdated (which in itself is interesting as it is only 12 years old.).
Reply
#5
Just finished the novel, took me a little longer than expected. It's a good book and I'd recommend it. Without spoiling anything it's about a world that is completely saturated with nanotechnology (like an angelnet) but the AI running it can no longer talk to or understand humans. Consequently the humans on this planet have fallen back to a dark age. The protagonist is a young mason who, in true hero of a thousand faces style, is thrust out of his simple life on a journey to understand what went wrong with the world, perhaps to fix it.
Reply
#6
(10-16-2013, 07:43 AM)Rynn Wrote: Interesting piece. I find the idea of automated science fascinating and a have often wondered if there will be a time in my life when a completely robotic team composes and tests a hypothesis. I have an amusing image of seeing a fully robot researched and written paper arrive on my desk and choosing that moment to start planning retirement :p

Here's an article about where we are in the timeline of automated science

http://m.cacm.acm.org/magazines/2012/5/1...y/fulltext

and two milestones that spell the small beginnings of that possibility.

The company Narrative Science has software that writes articles about your data.
http://narrativescience.com/
http://www.npr.org/2011/04/17/135471975/...s-reporter

Adam, created by British designers at Aberystwyth University in Wales
http://www.wired.com/wiredscience/2009/0...scientist/
Reply
#7
Automated science is going to be big in genetics research; we don't even know what to look for, yet, in many cases. A new way of looking at the way genetics is related to the abilities and health of an individual is needed; certainly the popular press gets it badly wrong. Maybe by removing human prejudices altogether we might begin to understand this stuff.
Reply
#8
(10-17-2013, 06:34 AM)Dfleymmes1134 Wrote:
(10-16-2013, 07:43 AM)Rynn Wrote: Interesting piece. I find the idea of automated science fascinating and a have often wondered if there will be a time in my life when a completely robotic team composes and tests a hypothesis. I have an amusing image of seeing a fully robot researched and written paper arrive on my desk and choosing that moment to start planning retirement :p

Here's an article about where we are in the timeline of automated science

http://m.cacm.acm.org/magazines/2012/5/1...y/fulltext

Thanks. It's interesting but I'm skeptical about the twenty year timeframe for biosciences. Having worked with physicists and engineers before who are sidelining into biology I'm very aware of how difficult biology is to model compared to those disciplines. It takes all sorts of scraps of knowledge and intuition to pursue a biology project whereas something more mathematically based would be simpler for machines.

(10-17-2013, 07:44 AM)stevebowers Wrote: Automated science is going to be big in genetics research; we don't even know what to look for, yet, in many cases. A new way of looking at the way genetics is related to the abilities and health of an individual is needed; certainly the popular press gets it badly wrong. Maybe by removing human prejudices altogether we might begin to understand this stuff.

I'm sorry I can't parse this, could you explain a bit more about what you mean?
Reply
#9
Every time a new relationship between a gene complex and a behavioural trait is even suggested by research, the media trumpet it as the discovery of a 'gene for intelligence' or a 'gene for gayness'. The reality is much more complex, but even geneticists are only just beginning to understand how complex this relationship is.

To make sense of the relationships between the genome and the health of individuals and the way they behave, we will need big data, big processors, and a way of conceiving and testing theories, and creating sophisticated models, rapidly and iteratively that is independent of human preconceptions. The same is true of research into the relationship between neurobiology and consciousness.

I know this is already starting to happen, but it is only the start of a process that will probably end by taking much of the donkey work of science out of human hands. We can't hope to replicate billions of years of evolution using pen and paper.
Reply
#10
That's media for you, shockingly bad and dangerous in it's faulty reporting. But biologists have know for a long time that the relationship between genotype and phenotype is hugely complex. Hence why all the omics have been formed, genes just aren't what scientists of old and the laymen of today think they are. A genome isn't a blueprint and a phenotype is as much a product of the environment as it is the genome (I don't just mean whether or not the organism lives in a forest or Ocean but the microenvironments within and outside of each cell).

If we had a lot of data on individuals genotypes and phenotype and got computers to search for patterns (which is what we do now) we can find interesting corelations but the hard part is finding what the causal links are if they are true corelations at all. I don't see where automated science can come in with that until software can understand biology as well as humans in which case we will already understand enough for this task.

Machine scientists would obviously have to work within the constraints of the models given to them and have the ability to test those models and develop on but model building is a lot more complicated and fuzzy in biology than other disciplines purely because the shear amount of variables to handle.[/quote]
Reply


Forum Jump:


Users browsing this thread: 2 Guest(s)