Replaced by A.I. ?

[rockstar.esq]

In the last week I've come across a couple of articles about how Artificial Intelligence (A.I.) programs replacing people.  One was written by a stock market analyst who surprised me by admitting that the "best practices" in use today aren't particularly good at making accurate predictions in today's market.   The author recommended that his colleagues switch  to a well-known technique that he claims works better.  He said they need to improve their predictions or they'll lose their jobs to A.I. programs.

Initially I thought it was remarkable that the people doing a specific job were using practices that they know don't actually work.  Then I got to thinking about Thomas Powers post about working at an ISO operation where following protocol is the primary objective no matter how counterproductive it may be.

The other line of work I read about was Human Resources.  Apparently the owner of a major international hotel chain is currently using A.I. to interview job applicants.  This one kinda shocked me because I expected the A.I. would be used only for drudgery like reading resumes or doing background checks.  

The owner of the company said his H.R. people were not answering applicant questions in interviews which was leading to applicants giving up on a job at his company. Given that we're talking about an international hotel chain, I think it's reasonable to believe that there are lots of managers who could/should/would be expected to correct this obvious problem.  With all that management in place, why spend more to develop a life-like A.I. program to interview applicants?

That leads to a more important consideration.

What if this hotel chain already has the "best" managers, and Human Resources personnel employing the very best of "Best Practices" for hiring new workers?  It's probable that these HR and management professionals have lots of very professional and established reasons to explain their behavior.  This implies their current work really is the best they will do.  This means the A.I. isn't being programmed to mimic the work of people it's replacing.  In this limited example, the A.I. is being programmed to be more courteous, respectful, and responsive than the people its replacing.  

This might be the cusp of something far more disruptive than just replacing repetitive-task workers with programs.  Entire vocations could be replaced because they've drifted from their purpose.  Like that financial analyst above, I suspect a lot of folks know  how to be more useful and efficient in their work.  This shifts the dynamic for job preservation considerably.  Stagnant professionals can't rely on job complexity, or even "the human touch" to shelter them from competition with A.I.  I was shocked to learn that there are already  A.I. programs  in development to replace therapists. 

The T.V. show "Frasier" was continually making light of  psychiatrists struggling with relationships in their own lives.  Looking at it from this perspective, it starts to take on a different meaning.  Frasier embodied a combination of arrogance and risk-aversion.  He was very proud of his intellectual and social abilities, yet he could never manage to apply himself honestly and sincerely to his pursuit until it was too late.  I can see a whole lot of people in my professional life who follow a similar pattern.  I doubt HR professionals thought their career would see competition from A.I. before the workers they oversee but that's what's happening.  

With all that said, which vocations do you think are next?

I think it will be Architects.

[JustAnotherViking]

Any thing which has regulations and well defined rules that can always be simplified down to yes or no, coupled with large data sets will be at risk. 

The ai still needs to be fed with large sample data sets to build predictive models before it can output an accurate enough result to replace a trained person. 

Architecture would have an artistic and design element which may take longer to model and be harder too adapt to changing trends. 

[TwistedCustoms]

I would say Engineering would be on the chopping block before Architects. I don't believe AI can ever be developed to the point it takes on an artistic sensibility. You can create a program to spit out designs that stay within a set of parameters such as Greek inspired proportions or Gothic whimsy but how do you give the software the ability or incentive to have a preference for one or the other. I haven't seen anything yet to make me believe true AI will ever materialize. I think the use of the term AI today is at best a misuse or misunderstanding of the term.

[ThomasPowers]

The ability to mimic human behavior seems to me to be based on size of databases and speed of processing; as such things increase then the better AI will work and the greater the number of tasks it will deal with.  I expect it to take a long long time to be able to handle impromptu stand up comedy! (On the other hand I have had some Drs visits where I would have done better with a current AI!) 

[Tubalcain2]

I saw this topic and it through me for a loop for a second. When you were raised on a farm AI has a bit of a different meaning.

[ThomasPowers]

My Daughter had a entire class on AI---she's a Vet.

[Charles R. Stevens]

Well, my first thought on reading the headline was...

we see artificial intelligence all the time, either the YouTube blacksmith or the Facebook award smith that comes here and is told no...

All kidding aside, human behavioral datransets are being mined as we sleek, Facebook and it’s competitors. As to artistic talent, we already have models of faces and objects humans find pleasing, it’s not a far leap for architecture. Besides have you seen the crap they have designed over the last 50 years? Sorry to any architects who are members but the average house plan today looks like it was designed buy one of my granddaughters with crayons

Psychological and medical “expert systems” (from baserkernfame) are not such a bad idea, human prejudiced is a thing, and if the programs are carefully designed to prevent this then better diagnostic outcomes will result. We have all spent time filling out medical questioners only to have them ignored and know at least one person misdiagnosed or treated with an outdated protocol. I don’t see doctors replaced in the near future but a lot less specialists might be in the horizon or at least patients being quickly screened so they go from GP to specialist surgeon with out weeks of delay and duplication  

[4elements]

I just saw a headline today, someone wants to create an AI deity

[SLAG]

Charles,

I agree with you completely.

There is an aphorism (tenet or saw), that computer programmers, systems operators/analysts, etc. say often.

"Garbage in garbage out".

That science is replete with good things and a surfeit of garbage, inane observations, rubbish. All of that generating, all too often, idiotic observations, and predictions.

I.f.i . stalwarts do not trade in your brain for an artificial Information System; it's a bad bargain and a poor substitute.

SLAG.

p.s. of course I am biased. We are all irreplaceable. 

[Charles R. Stevens]

Slag, the asylum is being ran buy the inmates! 7 billion peaple and the least common denominator seems to rise to the top... at least on my bad days.

[SLAG]

Herr Charles,

Oh my, you've been there?

I think we both graduated from the same asylum.

Regards,

The   SLAG.      (L.L.C.)

 

[JHCC]

14 hours ago, rockstar.esq said:

I was shocked to learn that there are already  A.I. programs  in development to replace therapists. 

Back in the '80s, someone was marketing the "Walk-A-Shrink", a cassette tape (remember those?) that you popped into your Walkman (remember those?) to play a deep reassuring voice saying, "Uh huh...uh huh...go on...right...and how did that make you feel?...right...tell me more...uh huh..." and so on.

10 hours ago, Charles R. Stevens said:

if the programs are carefully designed to prevent this then better diagnostic outcomes will result.

One thing computers are really good at is pattern analysis, and one area where pattern analysis is really, really important is reading EKGs. There was an article I read a little while back that described one of the first EKG-reading programs, which got tested against one of the best cardiologists in the business and came out far ahead.

[JustAnotherViking]

The current programs are good at pattern analysis when they are given a defined set of parameters to look for, and what to output.

The one thing they aren't even close to yet is the 'intelligence' part; being able to cross reference unrelated data sets without explicit instructions on what it actually represents, and what to do with it.

 

e.g., feed it thousands of hours of game data, it becomes a master of the game in minutes (go/chess).
ask it to apply those skills to an unrelated game and nothing happens.

It's just really fast and efficient statistical analysis, with less and less programming required to do the analysis part.

 

I'd say we're a very long way off a true AI, if such a thing is even possible.

[JHCC]

For now, let's stick to trying to convince people that Plaster-of-Paris and sand don't make a good refractory and that zinc and chrome platings shouldn't get burned off in a fire.

[TwistedCustoms]

33 minutes ago, JustAnotherBiker said:

The current programs are good at pattern analysis when they are given a defined set of parameters to look for, and what to output.

The one thing they aren't even close to yet is the 'intelligence' part; being able to cross reference unrelated data sets without explicit instructions on what it actually represents, and what to do with it.

 

e.g., feed it thousands of hours of game data, it becomes a master of the game in minutes (go/chess).
ask it to apply those skills to an unrelated game and nothing happens.

It's just really fast and efficient statistical analysis, with less and less programming required to do the analysis part.

 

I'd say we're a very long way off a true AI, if such a thing is even possible.

This!!! Yes!!! The "AI" we have seen so far isn't artificial at all, it's an extension of the intelligence of the person who programed it. At its core it's no more advanced than an automatic drill. If I set the limits on my drill press it can punch holes much faster than I can by hand but it's only performing a function that I "programed" it to do. If one night while I'm sleeping my drill comes up with a way to make drilling holes more efficient, designs and builds a new, novel device that does in fact do it better and that new device anticipates when and where I need holes drilled, and preempts the drilling of said tasks I will concede that AI has arrived. Till then "AI" is being used to describe machines that operate efficiently but in fact possess no intelligence.

[rockstar.esq]

 

15 hours ago, TwistedCustoms said:

I would say Engineering would be on the chopping block before Architects. I don't believe AI can ever be developed to the point it takes on an artistic sensibility. You can create a program to spit out designs that stay within a set of parameters such as Greek inspired proportions or Gothic whimsy but how do you give the software the ability or incentive to have a preference for one or the other. 

Twisted, your points about the artistic aspects are well made.  I've made something of an investigation into why Architects do what they do.  They're among the least satisfied people in their careers of any profession.  Their main complaint is that they spend the vast majority of their time in meetings about issues that have virtually nothing to do with aesthetics.  Architects are typically hired to serve as the owners representative for the purposes of bidding, contract award, and construction oversight.  I've looked at the Architecture course listings at Universities.  There's no requirement for them to take courses that cover; Construction law,  Accounting, Management, Scheduling, or Negotiations.  It's reprehensible how completely unprepared these people are for the actual job. I don't believe it's common knowledge but the vast majority of construction projects are  written using  templates created by the American Institute of Architects (AIA).  It's very noteworthy that the AIA has zero enforcement for design conventions or standards.  This makes it hard to tell what is incompetence and what is just inconsistency.

Beyond that, Architects assiduously avoid any liability for designs that exceed their clients budget.  This wouldn't be much of an issue if even 25% of them could design a project that met their clients budget.  Architects went from being gatekeepers for the clients and the builders, to a costly obstacle.  It often takes Architects longer to develop plans than it does to physically build the project.  Based on what crosses my desk, perhaps 30% of the "Final" plans make it through permit review without summary rejection.  If all of that wasn't bad enough, they allow specialty vendors to write specifications, and generate details that are passed off as the Architects work.  This inevitably limits competition, increases prices, and in the worst cases, feeds corruption.

Circling back to your original point, I think artistic/aesthetic photo palates like pinterest could be used to determine the client's preferences as well as the local aesthetics that are likely to be approved by planning and zoning.  It's not popular to point this out, but "thought leaders" in the Architecture profession have always been very vocal about how much they oppose designs incorporating function and conventionally attractive aesthetics.  They're pretty similar to the "high fashion" clothing designers on that score.  I've never seen anyone in real life who wore clothing like that, and I suspect that's half the point.  In contrast, insane amounts of public money is spent building the architectural equivalent of a sweater with nine sleeves.

Here in Denver, the city has become a laughingstock because of how ugly our architecture is.  A local Architecture critic for Westworld named it "McCentury Modern".  I like to explain it like this.  Corners are to buildings what elbows are to bodies.  "McCentury Modern" is what happens when the entire building is elbows.  Even prominent local architects have written newspaper articles agreeing that it's hideous.  Some go so far as to say that an aesthetic panel of judges needs to be implemented.  

My suggestion is less complicated.  Require that all buildings have a publicly visible placard with the name and contact information for the Architect of Record, General Contractor, and Zoning official who signed off on it.  If their work is good, they'll get repeat business.  If not, they'll hear about what they could improve.

"Till then "AI" is being used to describe machines that operate efficiently but in fact possess no intelligence."

I can 100% agree with your assessment here, and still see it as a realistic replacement for what most Architects are currently doing.

[TwistedCustoms]

rockstar.esq, Sadly I agree. The romanticized ideal of the Architect walking a property with a pipe clenched and eyes squinted, "seeing" something the rest of us can't see yet....well, that's not reality. I hate it that the professions that used to be based on artistic expression have been reduced to number crunching drudgery that can be performed by minor functionaries totally devoid of vision. If you are correct, and I suspect you are, that the technology being labeled AI today can replace the Architects of today then we are in a sad state of affairs.

[ThomasPowers]

I've worked in several buildings that won prestigious awards for their architecture that were Totally dysfunctional for working in! (Like the ESO headquarters in Garching, near Munich, Germany, where you could actually see where you needed to be across the atrium;  but could not figure out a way to get there!  Also the bathrooms had sliding "blast doors" like you would expect on a bunker.)  Or even the VLA building in Socorro.  The problem being trying to force offices into a structure that was "designed" to have a certain appearance.

[SLAG]

Disruptive technology will triumph over artificial intelligence very handily and that phenomenon is ineluctable (unavoidable).

It's non-linear.

An example is Kodak corp. and digital photography.

Artificial intelligence is, in the end, 'artificial'.

SLAG.

[TwistedCustoms]

21 minutes ago, ThomasPowers said:

 (Like the ESO headquarters in Garching, near Munich, Germany,

Part of their design considerations are dictated by the German law that workers can't be kept from Natural light for more than four hours at a time. Lots of glass in todays German Architecture!

[Charles R. Stevens]

Like all things it is advancing. Self learning architecture (AI programming) and Beowulf clusters have mad leaps for AI, as have folks realizing that AI doesn’t have to model human human intelligence or be all that intelligent. A lot of work has gonninto atonimus drones and other robots. At one point our ancestors weren’t much more intelligent than a cockroach. 

I here you saying computers are real good at patern recognition, so are humans. We recognize faces in just this way, we are so good at it in fact we see Jesus in potato chips. 

My best friends mother is (retired) an architect. I remember her taking pictures and taping them together in 360 so she could place the windows just so to catch the views. I also remember their home built on a slope facing desert vistas to the west. It was a nice livable house but for a kid who grew up in a trailer it felt like a movie theater. I came in the back door and stayed in my friends room as much as possible.

Mrs Helms wasn’t the only Architect I knew growing up, another friends Dad also was an architect. He was a bit more “grean” being that he was a 50’s beatnik and his wife a 60’s hippy, I didn’t get to see him at work but I did see some of his work. Their home was passive solar, earth remedy with brick floors and thick white plastered walls. A much more comforting space for a kid that grew up around Adobes. Tho for the life of me, every time Gavi invited me over for dinner it was samon and artichoke nite...

 

[JustAnotherViking]

Different sort of pattern recognition. 

A face in grilled cheese is a perfect example. Human will see a face, a program will see cheese because it can't cross reference in the way we can or spot the abstract patterns. 

It can however spot an emerging pattern in millions of flat record sets in seconds where as it would take a human years to analyse. 

An interesting field of study and relatively easy to get started in on a basic level with all the open source libraries. 

The key right now is the human understanding of the problem and converting that into a set of machine rules. 

In work I have access to a lot of useful raw data, but as I write the software rather than use it, my contextual knowledge is a bit too limited to start on the 'AI' programming in any useful capacity other than basic R&D/proof of concept modelling. 

[Charles R. Stevens]

I imagine the challenge faced at this point is taking the knowledge needed from the folks who have it and passing it on to programmers so they can write the programs. Some thing always gets lost in translation. 

But computer programs that learn and think are intelligent, even if to a limited degree. certainly not sentient but when you turn a car loose on city streets or a swarm of drones loose and they navigate a non lab inviroment they are exhibiting at least the degree of intelligence an insect has. Aren’t instincts just programs?  

I don’t think machine intelligence will ever be the same as human intelligence, we are driven buy some instincts that are simply of no use to every machine. Tho some machines may need to reproduce most wont, wile servival will be important for most machines some won’t benefit from that “instinct” a huge amount of human intellect is tied to herd/pack social interaction as a way to survive to pass on our genetic code. If we define intelligence as strictly human intelligence we miss the mark. An elaphant or a dolphin are intelligent, just nit human. Machines that make atonimas desks and desplay a very basic form of intelligence ( again the cockroach model)  

[rockstar.esq]

Lots of good points.  One thing to consider in all of this is that A.I. as it stands today isn't the point I was trying to make.  I think A.I. represents a potentially viable workaround for Entrepreneurs to bypass intractable professionals.  It requires significant investment and risk to early adopters, so it's going to be most worthwhile to replace professions that are especially costly and ineffective.

As individuals the "point" of all of this isn't to make doom and gloom predictions.  I really believe that successful people combined timing and opportunity.  Right now, there's a business pattern where a markets are bottlenecked to allow a specific (and often connected) vocation to serve as a gatekeeper to economic progress.

 This isn't the traditional "middleman" who's connecting supply with demand, this is actual supplier professions that shifted from competition into a "corporatist" approach to business where things are standardized to create barriers to entry and replacement.  They're not only in the way, there aren't any comparably skilled people who can help entrepreneurs get around them.

If a worried Architect was reading this, there's a huge opportunity for them if they'd consider some changes.  For starters, Architects could financially guarantee their performance as measured by total duration, quality standards delivered, and completed project cost.  Every individual thing I listed is what the client thinks they're getting when they hire an Architect.  I genuinely believe that any Architecture firm offering (and delivering) such a deal would utterly dominate their market.  I also believe that it's only a matter of time before the AIA attempts to curtail that trend because it threatens their current purpose.

Having said that, I think it's worth questioning whether the accepted "best practices" and the parties promoting them are actually aligned with the expectations of customers paying the bills.  I suspect that more often than not, they're at the root of "gatekeeper" behavior.  Corporatism has a huge target on it's back because entrepreneurs can (and will) spend less money building A.I. to permanently bypass them.

There are a lot of professionals who behave as though achieving a certification entitles them to a lifetime of career stagnation without competition.  Cleverly constructed obstacles to competition from other people will be the biggest incentive for entrepreneurs to invest in A.I.

I think there's a clear choice, either compete fully and faithfully to be worthwhile to your customer, or see your entire profession get bypassed.  

 

[Frosty]

What a great topic, thanks for starting it rockstar. I REALLY wish I remembered your name but that's okay this moniker is plenty unique if not personally satisfying for me.

Anyway, I see lots of valid observations and points so far. While I may not agree none are wrong IMNSOHO. 

Many years ago, whe the "learning(?)" channel actually was about education in a meaningful sense I watched a program about the bushmen of the Kalahari. What I found meaningful in an hour long and interesting program was the economics of living in the "probably" harshest inhabited place on Earth. The people were always migrating along traditional paths and they ate everything on the way. Everything within two paces unless it was something really good, (read large) any farther wasn't profitable. It cost more  energy and water than it yielded. This was the budget they've lived by fore ever. 

The endeavor MUST pay more than it costs.

This has colored my perspective since watching the episode. As a hobbyist smith it's not much of a factor, what little I usually sell I sell for materials and coffee money. If someone commissions me I bid based on $90/hr. it'd be more in the range of $150/hr if I were still active. In the mid '80s I charged $19.95 for a hook I could make start to finish in 7 minutes and people snapped them up as fast as I could make them at demos. I wasn't operating a business model though so they were going cheap.

Dad used to drag me to auctions and bored me to tears talking about bidding. His basic philosophy being: If you're winning all the bids you're too low if you aren't winning any you're too high or can't do the work. Same thing actually. Any business that makes a habit of either is who ended up at the auctions we went to. Dad taught me about bidding at auctions too, a good deal often wasn't. Cheap up front can be ruinous down the road.

So much for the rambling set up and back story. rockstar is intimately familiar with the topic, he deals with it professionally. All I have are observations based on my near constant reading and my almost manic free associating mind. 

AI, currently it's a marketing ploy. Computers aren't intelligent they're just very complex and have huge resources to access. Nor are the programs, computers are just the platforms, programs are what do things, the real machine cruising the roadmaps on the chips. So far computers aren't self aware or sentient, just good at doing what they're told. Unfortunately they're all too often told to do things based on a poor budget model.

Architectural design is something I don't have direct access too other than to see the results. BLEH! What I do have an intimate experience with is bridge design. A form of architecture with different requirements. I knew the bridge and road design guys by name and job, I used to be the draftsman in the same office till I got pulled to be the welder fabricator on the foundation drill crew and later a driller. Anyway, the guys almost never actually designed a bridge, they looked at the parameters, length, # of lanes and foundation data and tweaked the design from a book to match conditions.

Designing  bridges doesn't need a bridge engineer anymore just a manager to adjust the facts. It'd take longer to input the data than it'd take a computer to do the work. BUT reports MUST be written, checked, signed a couple times and filed. 

See the pattern I see? If something works file the design and use it till it no longer meets the parameters. When the base budget becomes the main parameter needless variations are bad things, things like creativity just make things harder to fit the model. 

I can never remember the guy's name and I dislike the entire theory strongly enough I won't search it out. He's the guy who thought up the theory of management. A manager doesn't need to know how to do the work to run the company. He reduced the business model to bean counting. Like any decision that is made with one condition carrying too much weight the over all mechanism has Darwined itself. Notice a change in your local super market since Kroger bought it out?

The local Safeway used to have one of the BEST in house bakeries around, better products and prices than the independants. However a local market chain  Carrs was bought out by first Safeway and then  Albertsons bought them out and now I think HAAGEN owns it all. A management company. Want to know what the Carrs bakery sells now? A limited variety of muffins, rolls, bread that comes as frozen pellets in foil pans or paper cups. In deference to honest advertising the goods are no longer said to be "made" here, they're BAKED here. In caps. Fred Meyer's new owner Kroger has a different line of frozen pellets that are BAKED FRESH!

The results after just a few years are new mom and pop bakeries opening, Krispy Cream opened one in Anchorage and from what I hear the pastries taste EXACTLY THE SAME as anywhere else. They're stupid too expensive for a raised donut so I haven't even tried one, even driving past on my way to town. If I cared I'd look and see who's management dictated frozen pellets are being BAKED FRESH at Krispy Cream but I don't.

With the over all control of business models falling to the bottom line flexibility fails, creativity becomes a BAD thing, choose 1 from column A and one from column B has become the norm for design in too many professions. 

Like any organism human productivity evolves, there is NO CHOICE, I don't care what Victorian thinkers say, we don't get to decide to stop evolution. What a laughable bit of hubris that one is. Like our Rockstar says quality architecture and other things are starting to be produced by small independants and they WILL eat the big bloated "established" organisms alive. The business worlds version of the Darwin Award. The auctions will be EPIC!

Frosty The Lucky.