SERVING THE QUANTITATIVE FINANCE COMMUNITY

 
User avatar
Cuchulainn
Topic Author
Posts: 59932
Joined: July 16th, 2004, 7:38 am
Location: Amsterdam
Contact:

Re: Three years from now: can you predict the Software Landscape in anno 2018?

June 10th, 2017, 4:07 pm

Image

Important areas indeed but somewhat mundane IMO. In the old days we called System Management, Procurement and Cost Control.. You wrote these systems in a language + OS + database.
What's different this time?
 
User avatar
outrun
Posts: 4573
Joined: April 29th, 2016, 1:40 pm

Re: Three years from now: can you predict the Software Landscape in anno 2018?

June 10th, 2017, 4:30 pm

I expect that companies decide to use AI based on a cost & performance metrics. They pick an AI if it's cheaper and/or better in any of these areas than humans or classical software.
 
User avatar
outrun
Posts: 4573
Joined: April 29th, 2016, 1:40 pm

Re: Three years from now: can you predict the Software Landscape in anno 2018?

June 10th, 2017, 5:16 pm

What's different this time is that we now have developed generic algorithms that can be taught to do certain tasks extremely well, these tasks were done poorly in the past. The new generic applicability is driven by lots of data + advances in statistical learning methods. 

A good example is voice synthesis. Stephen Hawking's voice synthesiser was build in the late 90s and we all know how that sounds. It required a team of people, a lot of time and domain knowlegde.

Last year google released paper about their WaveNet speech synthesiser. Two main differences stand out:
1) It was *trained* by proving it with speech samples using generic neural network methods. Any student can build this in a couple of months. The don't need domain knowledge.
2) is sounds ultra realistic, it has intonations, accents, emphasis on words, something that nobody was able to before.

.. so: much better performance + much easier to build -> everyone is using this now instead of the older methods.
 
User avatar
Traden4Alpha
Posts: 23951
Joined: September 20th, 2002, 8:30 pm

Re: Three years from now: can you predict the Software Landscape in anno 2018?

June 11th, 2017, 2:49 am

Image

Important areas indeed but somewhat mundane IMO. In the old days we called System Management, Procurement and Cost Control.. You wrote these systems in a language + OS + database.
What's different this time?
What's different is that it takes 90% fewer programmers.
 
User avatar
Cuchulainn
Topic Author
Posts: 59932
Joined: July 16th, 2004, 7:38 am
Location: Amsterdam
Contact:

Re: Three years from now: can you predict the Software Landscape in anno 2018?

June 11th, 2017, 9:22 am

What's different is that it takes 90% fewer programmers.

Maybe, maybe not. In any case, less white-collar jobs will be needed I suppose mainly due to automation in general. I don't know in how far "AI techniques" have spurred this on and how much is marketing from the big players.

The advent of AI will probably mean even more work for programmers + their communities as they maintain these new software system.

It takes maybe a whole lifetime to learn a natural language; if we ignore Google., Microsoft , and the other major players then it will take 10-15 years before the 'average' companies with limited resources and legacy systems make the transition, if at all. I reckon most of this stuff is in the laboratory.

Maybe I'm missing the point. Personally, I would _not_ like to work on many of the topics on that list (been there, done that under a different name and more scoped).

What was the question,again?
 
User avatar
Traden4Alpha
Posts: 23951
Joined: September 20th, 2002, 8:30 pm

Re: Three years from now: can you predict the Software Landscape in anno 2018?

June 11th, 2017, 12:49 pm

Yes, it's a maybe/maybe not situation. Each "mundane application" might require 90% fewer programers but the number of mundane applications written might surge and there's the added AI programmers at AI software vendors that you mentioned.

Yet the total will probably still drop in that 1 AI programmer at a cloud/SaaS/app vendor will replace 10 programmers at individual companies.

If one looks at the classic flow of data -> information -> knowledge -> decision, the old model required lots of white collar labor for every "->". AI reduces the labor at each link.
 
User avatar
Cuchulainn
Topic Author
Posts: 59932
Joined: July 16th, 2004, 7:38 am
Location: Amsterdam
Contact:

Re: Three years from now: can you predict the Software Landscape in anno 2018?

June 11th, 2017, 3:23 pm

Yes, it's a maybe/maybe not situation.  Each "mundane application" might require 90% fewer programers but the number of mundane applications written might surge and there's the added AI programmers at AI software vendors that you mentioned.

Yet the total will probably still drop in that 1 AI programmer at a cloud/SaaS/app vendor will replace 10 programmers at individual companies.

If one looks at the classic flow of data -> information -> knowledge -> decision, the old model required lots of white collar labor for every "->".  AI reduces the labor at each link.
Most of this can be automated? 
  1. Come into work
  2. Make coffee (and drink it!)
  3. Check email, reply to urgent ones
  4. Make coffee (and drink it!)
  5. Daily scrum meeting (tell the rest of the team what I've done yesterday, what I'm going to work on today, and voice any concerns)
  6. Meetings (design reviews, customer requirement gathering, etc.)
  7. Make coffee (and drink it!)
  8. Working on daily tasks (designing a system, or writing code, or reviewing someone else's code)
  9. Possibly shorter meetings to clarify some points/ask questions
  10. Go home
 
User avatar
Traden4Alpha
Posts: 23951
Joined: September 20th, 2002, 8:30 pm

Re: Three years from now: can you predict the Software Landscape in anno 2018?

June 11th, 2017, 3:37 pm

Yes, it's a maybe/maybe not situation.  Each "mundane application" might require 90% fewer programers but the number of mundane applications written might surge and there's the added AI programmers at AI software vendors that you mentioned.

Yet the total will probably still drop in that 1 AI programmer at a cloud/SaaS/app vendor will replace 10 programmers at individual companies.

If one looks at the classic flow of data -> information -> knowledge -> decision, the old model required lots of white collar labor for every "->".  AI reduces the labor at each link.
Most of this can be automated? 
  1. Come into work
  2. Make coffee (and drink it!)
  3. Check email, reply to urgent ones
  4. Make coffee (and drink it!)
  5. Daily scrum meeting (tell the rest of the team what I've done yesterday, what I'm going to work on today, and voice any concerns)
  6. Meetings (design reviews, customer requirement gathering, etc.)
  7. Make coffee (and drink it!)
  8. Working on daily tasks (designing a system, or writing code, or reviewing someone else's code)
  9. Possibly shorter meetings to clarify some points/ask questions
  10. Go home
The automation replicates the outcome, not the process. Aircraft produce flight even if that can't convert worms into flapping.
 
User avatar
Cuchulainn
Topic Author
Posts: 59932
Joined: July 16th, 2004, 7:38 am
Location: Amsterdam
Contact:

Re: Three years from now: can you predict the Software Landscape in anno 2018?

June 12th, 2017, 5:20 pm

The automation replicates the outcome, not the process. Aircraft produce flight even if that can't convert worms into flapping.

Could you train AI things to even dream of creating wing-flapping worms?
 
User avatar
Traden4Alpha
Posts: 23951
Joined: September 20th, 2002, 8:30 pm

Re: Three years from now: can you predict the Software Landscape in anno 2018?

June 12th, 2017, 5:32 pm

The automation replicates the outcome, not the process. Aircraft produce flight even if that can't convert worms into flapping.

Could you train AI things to even dream of creating wing-flapping worms?
Actually, YES!

There's been a lot of work in "artificial life" that uses AI techniques to adaptively evolve new lifeforms in a simulated environment. These simulations typically start with zero preconceptions about how the creatures should look but the system encodes the basic physics of the world. It's up to evolution to generate something that can move and find something to eat. Wing-flapping worms do arise.
 
User avatar
Cuchulainn
Topic Author
Posts: 59932
Joined: July 16th, 2004, 7:38 am
Location: Amsterdam
Contact:

Re: Three years from now: can you predict the Software Landscape in anno 2018?

June 12th, 2017, 5:59 pm

The automation replicates the outcome, not the process. Aircraft produce flight even if that can't convert worms into flapping.

Could you train AI things to even dream of creating wing-flapping worms?

At the end of the day I suppose AI algorithms are hard-coded big time?(?)
 
User avatar
Cuchulainn
Topic Author
Posts: 59932
Joined: July 16th, 2004, 7:38 am
Location: Amsterdam
Contact:

Re: Three years from now: can you predict the Software Landscape in anno 2018?

June 12th, 2017, 6:02 pm

About 15 years ago Agent Technology was a hot topic. Whatever happened?

https://en.wikipedia.org/wiki/Software_agent


Agents are the next major computing paradigm and will be pervasive in every market by the year 2000. (Janca, 1995)
 
User avatar
outrun
Posts: 4573
Joined: April 29th, 2016, 1:40 pm

Re: Three years from now: can you predict the Software Landscape in anno 2018?

June 12th, 2017, 6:15 pm

You have no idea hat you are talking about do you? Have you read things? Watched a lecture to see what's going on?
 
User avatar
outrun
Posts: 4573
Joined: April 29th, 2016, 1:40 pm

Re: Three years from now: can you predict the Software Landscape in anno 2018?

June 12th, 2017, 6:16 pm

About 15 years ago Agent Technology was a hot topic. Whatever happened?

https://en.wikipedia.org/wiki/Software_agent


Agents are the next major computing paradigm and will be pervasive in every market by the year 2000. (Janca, 1995)
Image
 
User avatar
Cuchulainn
Topic Author
Posts: 59932
Joined: July 16th, 2004, 7:38 am
Location: Amsterdam
Contact:

Re: Three years from now: can you predict the Software Landscape in anno 2018?

June 12th, 2017, 6:27 pm

You have no idea hat you are talking about do you? Have you read things? Watched a lecture to see what's going on?
So happens we created a distributed agent prototype in C# about 15 years ago at some expense. We were ready for more but after 9/11 the market imploded.

Enjoy your lectures. Maybe they told how to program BDI architectures.
ABOUT WILMOTT

PW by JB

Wilmott.com has been "Serving the Quantitative Finance Community" since 2001. Continued...


Twitter LinkedIn Instagram

JOBS BOARD

JOBS BOARD

Looking for a quant job, risk, algo trading,...? Browse jobs here...


GZIP: On