I'm a civil engineer and I don't see my job being replaced by a computer anytime soon. The tools we use (AutoCAD) will get much better, making my job easier and give me more options when performing a design, but it won't replace me. There are just too many variables for a computer to do the work, varying from the goals of the client to the individual characteristic to each site. One thing that I learned while getting my degree is that once physics starts dealing with the real world it gets incredible complicated, and computers give us the opportunity to eliminate some of that complication, but it still needs a human hand to guide and do the critical thinking about how all the variables get adjusted to get the most accurate solution.
I don't think the field I've worked, marketing, in will see automation ever fully take over. One part because it relies on human actions of the consumer and predicting this, as illustrated by the election, is far from exact. Most advancement in software has really just been in quantifying the results. It makes budgeting and managing easier but far from replacing human decision making in the process totally. Second is the creative side of marketing which may never be replicatible by computers. Producing marketing materials in the forms of copy, advertising, etc on a truly creative level is pretty safe from automation since it deals with replicating human emotions and concepts on a very cognitive level. I think it was Trakiel that brought it up in the other thread, the idea of a post want/need society. This is one subject Im actually optimistic about while other see doom. The argument going that with huge swaths out of work we'll see social strife lead to violence. We will need to bridge the gap between the time most manual labor jobs are automated and when we advane enough to find a reliable limitless energy source, where the idea of universal basic income comes in. I like to think more along the lines of Star Trek's positivity for human kind. Where people's drive is in bettering their knowledge and artistic endeavours instead of being defined by a life time of providing from a career.
This is going to be very difficult for the US especially because we're an extremely work-oriented culture. I feel like the finances of a post-work society will be easier to figure out than the cultural shift it will require.
Yeah, the cultural shift will be a sticking point. Why I mentioned the energy issue is because once we are able to produce unlimited cheap energy it takes away the need for human productivety and in the more fantastical Star Trek prognostication, lead to producing most of our needs with matter replicators. It would really force us to adapt socially/culturally to a different paradigm. Advancing our own personal knowledge or artistic abilities would become the outlet for hard work. Star Trek was very utopian in this area, something I like to hope we can get to. Getting into more of the sci fi weeds: The other idea I love contemplating is the advancement in technology in making knowledge universally availible cybernetically, ala The Matrix style uploading directly into people's minds. Imagine at some point where a singularity level technology pushes computers past human capabilties. Theoretically we'd be able to intergrate this back and forth. Everyone would have access to all of human knowledge and the there would be no gap between the rich and poor in terms of education and cognative abilities. Imagine a society where instead of schooling, knowledge is uploaded into everyone's mind at some point in their life. How radically different we'd become if everyone's baseline is all of human knowledge.
They should keep me around. I work in a luxury field that is equal parts science and art. Sure, you can machine harvest and mass produce grapes for mediocre wine, but the higher quality stuff, the stuff people will pay $100/bottle for? That is hand farmed, hand harvested, and the skill and intuition to know when to do that, as well as how to blend things for the correct flavor profiles, that's all a mix of numbers and art.
Kind of a far stretch to see rich humans liquidating other humans for marginal inconvenience. You really have to worry about those robots with their cold calculating nature.
You're completely delusional. Every single part of what you do could be boiled down to measurements and optimized. There's no "art" in wine making, only things that you don't have tools and machines to measure. One of the biggest things that people miss with AI is that learning and development will start increasing at an exponential pace. Winemaking, or any kind of alcohol/food field, is basically just optimizing crops and flavors. People call it art just because they don't have the right science to execute every piece of the process, but "educated guessing" doesn't make it art. It's just a poorly optimized science. And I say this as a lover of fine wines.
To this point, you may see a fair bit of reporting on this thing called "Machine Learning". It's what brings the decision-making and "artistry" to artificial intelligence. It's the process by which a machine can (for example) process 10,000 classical songs and figure out what makes it good or bad, and then go create something new. It's the core of self-driving cars, among a whole bunch of other things, and it is a HUGE topic right now. Corporations are already utilizing it to learn from that shit-ton of data that they collect from you, and the service is already readily available to the general public: https://cloud.google.com/products/machine-learning/ Hell, Google pre-seeds a lot of their stuff with huge information packs. Tesla has over 1.3 billion miles of roads travelled in it's "learning database" for their autopilot. Like any technology, now that it's readily available to the "common" person, it will fllourish and will be used to do things nobody thought possible 5 years ago.
I read an interesting piece that argued that the era of "easy tech" is over. By that they meant that the days of creating an app or a website and making a ton of money are (mostly) over, and that all the gold left in them thar hills is in the really hard problems like machine learning. They compared Google's Introduction to Android tutorial with their Introduction to TensorFlow tutorial and just pointed out the magnitude step up in complexity and assumed intelligence of the developer between the two.
Ill plead ignorance on "machine learning" but when it comes to art how would computer programing make decisions on something that comes down to subjective taste? Painters or wine makers that are truly great have the tools, proper equipement and training, plus that creative spark which admittedly Im not sure how you'd be able to replicate without some sort of sentient AI*. It's just beyond me. * All of this explored in Star Trek TNG with Data and his quest to be more human.
True "deep" art will never be replaced, because its goal is to constantly be wholly original. "Artisanal" however, is mostly just a function of deep knowledge which can and will be emulated by computers. Wine is not art, it is artisanal.
The bigger point is that as computers and programs get more sophisticated we're coming to understand that a lot of things which we once thought were purely subjective can in fact be predicted, modeled, and replicated.
It's in the name... "Learning". Those painters didn't come out of the womb with their abilities, they were learned. ML is about trying to build a process and then feed that process with data that allows the machine to learn that information so that it can have its own experiences in order to make similar judgments.
Exactly. This isn't a subtle 3 hour movie with a complex plot (which may be able to be modeled anyway, but is a different level of complexity). This is a set of flavors with a finite set of variables that contribute to them. Sugar content of the grapes, chemicals from the skins, fermentation products... there are a handful of inputs that create a finished product. The fact that wine makers don't have a small tool that instantly reads the chemicals and sugars and flavor profile of each bunch of grapes (or each barrel of wine to blend) and provides a complete analysis means that experienced farmers have to use their judgement, and they call it art. But that's not art, that's just imperfect science. It's like saying the internal combustion engine is art just because you don't have the tools to test compression, so you use an educated guess and some manufacturers are better at it than others.
But it goes further than that... they are actually learning how to produce art. It's not just about following a cookbook using a mass spectrometer for an input... that is "easily" done now with current technologies. The end goal is about neural networks and actually learning. For instance, Google's Deep Dream... it's a rough foundation that is being taken and utilized in other visual/artistic areas, such as painting. And music. The science is definitely in its infancy, and it has a long way to go, but the recent advances are staggering.
Totally agree. Art is not really as subjective as many would like to believe and we're making staggering progress on identifying exactly what makes it good. All I'm saying is that a lot of fields that are currently thought, by people in them, to be "art" are nothing more than shitty science that rely on deeply experienced people to make up the gaps that empirical measurements haven't filled in. Human brains are pretty good at pattern recognition and can put together things like "grape color + leaf color + seasonal temperatures this year + flavor + the look of the rest of the grapes in the field = good harvest time." But that doesn't mean a machine couldn't do it better given the right tools.
I just spent the last two hours getting a full education on Irish whiskey at the Bushmill distillery in Ireland. So fuck wine. (I still love you, wine)
Absolutely. I've been looking into "Big Data" and Machine Learning for about a year now, and my current project is about doing just that (basically finding results in data lakes that you don't specifically go looking for or program for), and it's really, really interesting stuff. Your AI results are only as good as your model (or rules that are applied). We are now at a stage in AI or Machine Learning where we are not having to actually define the model any more... the computer is self-learning that model. That's what Machine Learning is... it's basically, "here's a bunch of information, process it and come up with some findings on your own." You can direct it to find weird data associations in online purchases, or drug interactions in pharma/Medicare data, or any other bunch of large data. And that's really what the human brain has been doing your entire life... collecting data from your senses, and then building a model on how to interpret that data. The parallels are kind of freaky.