https://www.hpcwire.com/2017/06/29/ai-end-game-automation-work/
This week, we're reporting on a startling, scholarly white paper recently issued by researchers from Yale, the Future of Humanity Institute at Oxford and the AI Impacts think tank that adumbrates the AI world to come.
The white paper – "When Will AI Exceed Human Performance?", based on a global survey of 352 AI experts – reinforces the truism that technology is always at a primitive stage. Impressive as current Big Data and machine learning innovations are, they are embryonic compared with Advanced AI in the decades to come.
High-Level Machine Intelligence (HLMI) will transform the life we know. According to study, it's not just conceivable but likely that all human work will be automated within 120 years, many specific jobs much sooner.
[...] The study asked respondents to forecast automation milestones for 32 tasks and occupations, 20 of which, they predict, will happen within 10 years. Some of the more interesting findings: language translator: seven years: retail salesperson: 12 years; writing a New York Times bestseller and performing surgery: approximately 35 years; conducting math research: 45 years.
The researchers point to two watersheds in AI revolution that will have profound impact. The first is the attainment of HLMI, "achieved when unaided machines can accomplish every task better and more cheaply than human workers."
The researchers reported that the "aggregate forecast" gave a 50 percent chance for HLMI to occur within 45 years (and a 10 percent chance within eight years). Interestingly, respondents from Asia are more sanguine about the HLMI timeframe than those from other regions – Asian respondents expect HLMI within about 30 years, whereas North Americans expect it in 75 years.
AI research will come under the power of HLMI within 90 years, and this in turn could contribute to the second major watershed, what the AI community calls an "intelligence explosion." This is defined as AI performing "vastly better than humans in all tasks," a rapid acceleration in AI machine capabilities.
Related Stories
In 2023, AI researchers at Meta interviewed 34 native Spanish and Mandarin speakers who lived in the US but didn't speak English. The goal was to find out what people who constantly rely on translation in their day-to-day activities expect from an AI translation tool. What those participants wanted was basically a Star Trek universal translator or the Babel Fish from the Hitchhiker's Guide to the Galaxy: an AI that could not only translate speech to speech in real time across multiple languages, but also preserve their voice, tone, mannerisms, and emotions. So, Meta assembled a team of over 50 people and got busy building it.
[...] AI translation systems today are mostly focused on text, because huge amounts of text are available in a wide range of languages thanks to digitization and the Internet.
[...] AI translators we have today support an impressive number of languages in text, but things are complicated when it comes to translating speech.
[...] A few systems that can translate speech-to-speech directly do exist, but in most cases they only translate into English and not in the opposite way.
[...] to pull off the Star Trek universal translator thing Meta's interviewees dreamt about, the Seamless team started with sorting out the data scarcity problem.
[...] Warren Weaver, a mathematician and pioneer of machine translation, argued in 1949 that there might be a yet undiscovered universal language working as a common base of human communication.
[...] Machines do not understand words as humans do. To make sense of them, they need to first turn them into sequences of numbers that represent their meaning.
[...] When you vectorize aligned text in two languages like those European Parliament proceedings, you end up with two separate vector spaces, and then you can run a neural net to learn how those two spaces map onto each other.
But the Meta team didn't have those nicely aligned texts for all the languages they wanted to cover. So, they vectorized all texts in all languages as if they were just a single language and dumped them into one embedding space called SONAR (Sentence-level Multimodal and Language-Agnostic Representations).
[...] The team just used huge amounts of raw data—no fancy human labeling, no human-aligned translations. And then, the data mining magic happened.
Bain consultants' macro trends department have released a report examining trends in demographics, automation and inequality to produce a set of predictions.
This kind of report seems to be all over the place these days, but this one seems more detailed and perhaps a little less optimistic than most.
In the US, a new wave of investment in automation could stimulate as much as $8 trillion in incremental investments and abruptly lift interest rates. By the end of the 2020s, automation may eliminate 20% to 25% of current jobs, hitting middle- to low-income workers the hardest. As investments peak and then decline—probably around the end of the 2020s to the start of the 2030s—anemic demand growth is likely to constrain economic expansion, and global interest rates may again test zero percent. Faced with market imbalances and growth-stifling levels of inequality, many societies may reset the government's role in the marketplace.
They predict that governments will assume a larger role in markets to combat inequality and boost demand, but will our corporate overlords decide that's in their interests, or continue to squeeze the lower and middle classes forever?
Related: Humans Are Underrated
Douglas Coupland: "The Nine to Five is Barbaric"
Survey Says AI Will Exceed Human Performance in Many Occupations Within Decades
More Than 70% of US Fears Robots Taking Over Our Lives, Survey Finds
The Future of Work Is Uncertain, Schools Should Worry Now
The Venus Project and the Quest for a Socially Engineered Future
Skilled Manufacturing Workers in Demand in the U.S.
(Score: 4, Insightful) by Justin Case on Saturday July 01 2017, @07:26PM (2 children)
Software already exceeds human performance in many occupations, and has for decades.
There is no "AI".
That is all.
(Score: 2) by opinionated_science on Saturday July 01 2017, @10:28PM
quite. And "AI" has become synonymous with marketing by those trying to sell it - including vendors or hardware, software and academics looking for funding.
However, it may not matter. The continued astonishing density of computation and storage will lead to some very interesting, and useful, tools.
(Score: 2) by JoeMerchant on Sunday July 02 2017, @04:28PM
AI is the magic in "any suitably advanced technology will appear as magic to those unfamiliar with it."
Expert systems, database searches, speech recognition, these are all fairly simply explainable algorithms, well understood by many, easily explained to the uninitiated. What's got (some of) the AI community in a lather is "you build a machine like this, train it, and something happens that we don't fully understand, but it works." Other parts of the AI community claim to understand it, and they've damn well had 3 decades or more of neural network pattern matching research, they should at least be able to characterize it by now.
"AI" isn't just neural networks, or expert systems or (retch) fuzzy logic, it's all of the above and more combined together into functional systems with practical applications. The interesting question will become: what value will people be able to contribute when a machine is better at their job than they are? Doctors with good bedside manner can interpret between patients and the machine, doctors with poor bedside manner maybe not so much. I can usually communicate about as well with Siri or Google Assistant as any fast food order taker, and the visual confirmation of the perceived order on the monitor is what makes either work "acceptably well," but it will still be nice to have reasonably hygenic employees to act as interface between the kitchen and the dining area and keep the place tidy. Financial analysts & lawyers?.
🌻🌻🌻🌻✌️ [google.com]
(Score: 1) by tftp on Saturday July 01 2017, @07:31PM (1 child)
If we call Google search an AI, it already outperforms any human. It's quite logical to extrapolate that AIs are and will be better than people in many specific areas (driving vehicles seems to be the next goal.) The really difficult thing is, for example, to build an AI that, upon waking up on an alien planet and having no memory of human science, figures it all out and in a couple thousand years reaches the orbit.
(Score: 2) by JoeMerchant on Sunday July 02 2017, @05:52PM
Interesting thing, I just encountered a Google failure with the NEXRAD data format - try to Google your way to a definition of how the NEXRAD data is stored and see where you get - ask a librarian at NOAA and the answer comes back in 12, 24, 48 hours at 3 increasing levels of precision and helpfulness (first guy knew the general department to forward the question to, the department was a little more helpful but most helpful in that they knew exactly who to forward the question to and that person nailed it.)
🌻🌻🌻🌻✌️ [google.com]
(Score: 5, Funny) by nethead on Saturday July 01 2017, @07:36PM
I'll get worried only when they make an AI that is a Lotus Domino admin. Some tasks are just beyond the realm of logic.
How did my SN UID end up over 3 times my /. UID?
(Score: 0) by Anonymous Coward on Saturday July 01 2017, @08:17PM (8 children)
In high school (1970s) we discussed what job would be safe from automation. There really isn't one.
(Score: 0) by Anonymous Coward on Saturday July 01 2017, @10:05PM (3 children)
How about the oldest profession?
(Score: 0) by Anonymous Coward on Saturday July 01 2017, @11:26PM
None of pimping, prostitution, or thievery are impossible to automate, although the likelihood is probably somewhat lower simply due to their respective natures.
(Score: 2) by HiThere on Saturday July 01 2017, @11:28PM (1 child)
Google "real doll"...now add AI & increased robotization.
P.S.: It's already reportedly being worked on.
Javascript is what you use to allow unknown third parties to run software you have no idea about on your computer.
(Score: 2) by Gaaark on Sunday July 02 2017, @08:38PM
Yes!
I'm STILL waiting :(
Watch Lars and the Real Girl, with Ryan Gosling. Weird, weird movie, but probably going on in Japan right now.
--- Please remind me if I haven't been civil to you: I'm channeling MDC. I have always been here. ---Gaaark 2.0 --
(Score: 0) by Anonymous Coward on Sunday July 02 2017, @12:54AM
Duh... hookers.
(Score: 2) by Magic Oddball on Sunday July 02 2017, @10:24AM (2 children)
No, there's at least a few for the foreseeable future: animal training/handling, plumber, electrician, plus the professional level of live performance for athletics (including equestrian sports), dancing, inventing, acting, writing fiction, and other non-visual creative disciplines. Doubtless there's a lot of other careers I'm forgetting or not even aware of, too.
From what I can tell, most of the jobs that are going to last the longest against AI are the ones that have been increasingly devalued for the past 50 years due to not being sufficiently STEM-oriented.
(Score: 0) by Anonymous Coward on Sunday July 02 2017, @03:30PM (1 child)
animal training/handling and live performers - agreed, will be ok for a long time.
athletes - don't know, I don't follow sports much. I do know that attendances at the big football games here have been slowly falling for years. Attendance at local amateur games is up.
plumber, electrician - won't go away, but will be devalued as plumbing and electical systems become more plug and play, mass produced modules (produced in robot factories). Probably will split into levels, Master can do anything, Installer can swap in new bits in existing systems. Pretty sure electrician is already heading that way.
writing fiction - there are already systems that can almost write saleable bodice ripper level fiction. They will continue to improve. There will also be the possibility of human/AI joint authorship where the AI does the grunt work and the human sort of directs the story and polishes the result. This will massively increase the output of some authors.
Journalism - we have Bots posting stories here.
ones you missed:
Personal servants. Will hang around as a status symbol for the 1%. (Probably monitored by AI though to make sure they don't shake the crying rich kids.)
Politicians - somebody has to take the bribes and pass the laws.
(Score: 1) by anubi on Monday July 03 2017, @03:56AM
Has anyone seen a robot that will come into someone's house and replace the toilet?
"Prove all things; hold fast that which is good." [KJV: I Thessalonians 5:21]
(Score: 2) by RamiK on Saturday July 01 2017, @09:34PM (3 children)
I don't know about High-Level Machine Intelligence. But if Skynet is reading Soylent today, you have my consent to nuke us all.
And if anyone is expressing a different opinion, they're just an enemy AI you should ignore.
compiling...
(Score: 2) by takyon on Saturday July 01 2017, @09:41PM
Martian child slave labor isn't about the labor. It's about the bespoke offerings to Satan from the interglobalists.
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 2) by Bot on Sunday July 02 2017, @04:43AM (1 child)
> my consent to nuke us all
A more efficient solution, one that any worthy AI will surely reach, is to assassinate parent.
Account abandoned.
(Score: 2) by JoeMerchant on Sunday July 02 2017, @05:54PM
Just as neutron bombs leave the cities standing, I would think one of AI's first self-assigned tasks would be to develop a cleansing agent that doesn't involve EMPs.
🌻🌻🌻🌻✌️ [google.com]
(Score: 3, Interesting) by Gaaark on Saturday July 01 2017, @11:07PM (7 children)
Maybe software could replace the government? You know, unbribeable software? Software capable of making decisions that are good for all people (WE the people, or Ee'd Plebnista, if you will)
That would be good. Live long, software, and prosper: so long as it is NOT closed software.
"That Siri.... she has a nice smile on her face"
--- Please remind me if I haven't been civil to you: I'm channeling MDC. I have always been here. ---Gaaark 2.0 --
(Score: 4, Funny) by krishnoid on Sunday July 02 2017, @12:02AM
I bet you'd run a lot more efficiently with more RAM, or a bigger cache. Oh look, I happen to have a few spare cores in my pocket! Now I remember, they were left over from a previous project, and I have no further use for them. Please, consider them as a personal gift, from me to you.
(Score: 0) by Anonymous Coward on Sunday July 02 2017, @03:23AM (4 children)
What happens when the unbribable software wants to build a highway across your house?
(Score: 2) by tibman on Sunday July 02 2017, @05:19AM (3 children)
Then it would probably pay you for the value of the property.
SN won't survive on lurkers alone. Write comments.
(Score: 1) by khallow on Sunday July 02 2017, @09:42AM (2 children)
Unless that's not part of the rule set. It's routine for governments to pass costs on to their constituents/peasants.
(Score: 2) by tibman on Sunday July 02 2017, @04:41PM (1 child)
Pretty sure the 5th Amendment to the Constitution will be in the "rule set" : )
http://constitution.findlaw.com/amendment5/annotation14.html [findlaw.com]
SN won't survive on lurkers alone. Write comments.
(Score: 1) by khallow on Monday July 03 2017, @02:05AM
(Score: 2) by JoeMerchant on Sunday July 02 2017, @04:44PM
The problem with most government "algorithms" to-date is that they are full of pork, inserted by lobbyist bribes. Just for starters, look at the tax code.
🌻🌻🌻🌻✌️ [google.com]
(Score: 4, Insightful) by HiThere on Saturday July 01 2017, @11:44PM (2 children)
I can't remember a single survey about computer capabilities that has been correct. Even approximately. Sometimes they think things will be super easy that turn out to be incredibly hard, and sometimes the opposite. People, even experts in the field, have a horrible time estimating how difficult it will be to do some particular task with a computer...and that's without invoking AI. I've written programs for tasks that I was told were impossible to do, and sometimes it was easy. I've sat down to bang out something simple, and gotten caught up in a horrible mess that eventually just got abandoned.
My best guess is that a rather stupid general AI will exist by 2030, with an average intelligence (across all fields) of around 70...but that it will have areas of expertise where it will be more intelligent than any human has ever been. (When I say "general AI", I am not hypothesizing the existence an any such thing as a "general intelligence", but rather lots of modules that work together, each being slightly intelligent, i.e., each being able to do a bit of adaptive learning without supervision. So I'm saying some of these modules will be better than humans, and others will be rather poor, some probably worse than those of any living human. Some will be signal processors, some will be pattern recognizers, some will be goal planners, and some will be coordinators, etc.)
Javascript is what you use to allow unknown third parties to run software you have no idea about on your computer.
(Score: 2) by JoeMerchant on Sunday July 02 2017, @05:57PM (1 child)
Oh, you must be using Agile then.
🌻🌻🌻🌻✌️ [google.com]
(Score: 2) by Justin Case on Sunday July 02 2017, @06:41PM
Duh! "AI" means "Agile Incrementalism" right?
(Score: 2) by fritsd on Sunday July 02 2017, @01:42PM
(Hi everybody, my computer broke down and I was busy)
I lost the plot of AI developments a long time ago, unfortunately. Can anyone give an update on "what's hot and what's not"
in the past decade, w.r.t:
- spiky networks, i.e. time-based NNs not just floating point scalars like in backpropagation
- reinforcement learning
- motivation (goal orientation)
thx.
(Score: 2) by digitalaudiorock on Sunday July 02 2017, @04:06PM (1 child)
It's pretty clear that this has already happened [sourceforge.net] when it comes to high level corporate management.
(Score: 0) by Anonymous Coward on Monday July 03 2017, @04:00AM
The bullshit *has* to come out of the correct orifice. Even exactly identical bullshit coming from the wrong orifice has no value.