The Liberal Gun Club Forum

Welp. Let's not give this AI the keys to anything.

https://www.iflscience.com/technology/a ... sly-wrong/

Try it here:
http://inspirobot.me/

My first go with Inspiro:

Awesomeness! I say givem just enough rope... Maybe the developers of AI will discover just how bad an idea it is for us to be playing GOD at our stage of emotional development. That online AI experiment where the robot essentially turned into a racist bigot should have been extremely telling.

On my second or third try,

it seems to think I'm a republican or something...

Well played inspirobot.

Is IFLScience still a controversial site? Charges of stealing, plagiarism, click bait, hyperbole.

Found this article from 2015.

https://medcitynews.com/2015/08/iflscie ... care-news/

As a fan of goth music I think Inspirobot and I could be good friends.

Bisbee wrote: Wed Nov 14, 2018 2:08 pm Awesomeness! I say givem just enough rope... Maybe the developers of AI will discover just how bad an idea it is for us to be playing GOD at our stage of emotional development. That online AI experiment where the robot essentially turned into a racist bigot should have been extremely telling.

I am reminded of the1960s SciFi sort story about the computer that was asked the question, "Is there a God". The story goes on as the computer science people build bigger and faster computers. Finally they ask the question again and the computer checking to make sure its power supply is protected in its control says, "There is now."

BKinzey wrote: Thu Nov 15, 2018 12:35 am Is IFLScience still a controversial site? Charges of stealing, plagiarism, click bait, hyperbole.

Found this article from 2015.

https://medcitynews.com/2015/08/iflscie ... care-news/

Click bait and hyperbole is a staple there.

Uh, inspirational?

It's definitely not intended to be inspirational.

shinzen wrote: Fri Nov 16, 2018 11:14 am It's definitely not intended to be inspirational.

Randomly sardonic?

Reminds me of Despair.com stuff

[new_window][/new_window]

Eris wrote: Fri Nov 16, 2018 12:35 pm Reminds me of Despair.com stuff

That was one I had never heard of, so I checked it out. You're right - here's an example:



Grim stuff, but most of it seems pretty funny - either that, or I'm just a sick puppy. Probably the latter...

joemac wrote: Sat Nov 17, 2018 3:07 pm [new_window][/new_window]

Eris wrote: Fri Nov 16, 2018 12:35 pm Reminds me of Despair.com stuff

That was one I had never heard of, so I checked it out. You're right - here's an example:

Grim stuff, but most of it seems pretty funny - either that, or I'm just a sick puppy. Probably the latter...

I ordered a pack of cards from them recently and gave them out at work to demotivate people.

Eris wrote: Sat Nov 17, 2018 8:06 pm
I ordered a pack of cards from them recently and gave them out at work to demotivate people.

Popular ‘round the office are you?

Yeah, these should be great water-cooler conversation starters...like this this one:



Maybe we need to merge this with Hucks yuks thread. Eris - you made my Christmas gift buying way easier this year. They've got mugs...

Bisbee wrote: Sat Nov 17, 2018 8:34 pm Popular ‘round the office are you?

I'm a programmer. Sarcasm is a fundamental skill in my industry.

Bisbee said:

Maybe the developers of AI will discover just how bad an idea it is for us to be playing GOD at our stage of emotional development.

Some of the tools my team and I use look a lot like AI, but they are closer to machine learning - i.e., they discover stuff, but are seldom used to take action on the discovery - there is a human intermediary step.

The most recent Journal of the ACM published the updated "Code of Ethics" for computational scientists and programmers: https://www.acm.org/code-of-ethics

I am a lifetime member of both the ACM and the IEEE. Without trying to start an inter-society rivalry, the ACM is the more academic, thinking-persons society, while the IEEE is a bit more "Morlock" in nature.

The codes-of-ethics for both organizations emphasize only working on projects where one is actually qualified, but that is a self-referential definition lacking in outside oversight. The deep AI papers in the ACM have something similar to Asimov's Laws-of-Robotics. The most important of which is that we generally leave autonomous systems only loosely coupled.

There is a great concern amongst computer scientists about autonomous vehicles because they will get the best results if they: (1) Share a lot of information; (2) Make interpolations and extrapolations based on data patterns; (3) Reshare new algorithms and decision trees; (4) Cross share statistical information about the past behaviors of nearby human drivers.

As Bisbee intimated, that massive cross sharing may be a technology "ahead" of our societal, legal and emotional development. But it will happen - and largely without our ability to control what is shared or where and with what/whom.

All of this is a bit above my pay grade. I use AI and ML with data and add in a lot of mathematical (generally not brute force) analytics. My branch of mathematics is more useful for analysis than "prediction", etc., so I tend to be removed from most of the deep learning and autonomous systems. I will say one thing, and I mean no offense: many of the AI folks do speak of themselves in "god-like" terms. They are far more impressed (in general) with their progress than their shortcomings.

Here is a major change since I started working on systems in the early 1980s: "Frameworks". In 1985, when I released a large scale communications platform, my team and I had written 99% of the code that was in the system. We used an early "open source" ISAM code base for some of the data management. The rest, we wrote by hand. Inefficient for sure, but you knew what every line of code did.

That is all gone. In some cases, modern systems actually have a reverse ratio - the released systems have so many frameworks in them that the "developers" have written 1% of the code they deployed. The other 99% was written by someone else - and in spite of the open source fetish for "community inspection", we have some evidence that many frameworks are put into use without much anlaysis beyond reading the purpose and the API. The only real testing seems to be for performance and memory leakage. I have never seen a development team do deep code reviews for the frameworks they use - it may be happening somewhere, but I have never seen it.

So what to make of all this? Watch the Disney movie "Fantasia" and see the Sorcerer's Apprentice segment. Experimental thinkers have been unleashing new technology on society for centuries without understanding the consequences. I am sorry, but I don't think my point of view is either pessimism or nihilism; methinks it's called 'realism'.

Yup, and now to our correspondent in Beijing for other developments to creep you out...
https://youtu.be/eB29ZVDOFfU

My personal favorite from years of working in corporate America, and certainly applicable to our major Politico Parties.