nanog mailing list archives

Previous By Date Next
Previous By Thread Next

Re: The Real AI Threat?

From: Miles Fidelman <mfidelman () meetinghouse net>
Date: Fri, 11 Dec 2020 12:22:27 -0500
Um... there are long standing techniques for programs to tune themselves & their algorithms - with languages that are particularly good for treating code as data (e.g., LISP - the grand daddy of AI languages - for whatever definition of AI you want to use).
And... a common complaint with current machine learning algorithms, is that they often "learn" to make decisions that can't be understood, after-the-fact.  We already have examples of "racist bots," and there are lots of legal issues regarding things like liability for injuries caused by self-guiding cars.
And then there are "spelling correctors" and digital "assistants" - when has Siri EVER done only what you want "her" to do?
The REAL problem is programs that blindly go off and do what you think you told them to do, and get it woefully wrong.  The more leeway we allow our programs to adapt, or learn, or self-tune, or whatever-you-want-to-call-it - the more trouble we're in.
(The point being:  We don't have to wait for "real" AI to see many of the dangers that folks fictionalize about - we are already seeing those dangers from mundane software - and it's only going to get worse while people are looking elsewhere.) 

Miles Fidelman

J. Hellenthal wrote:
Let me know when a program will rewrite itself and add its own features ... then we may have a problem... otherwise they only do what you want them to do. 

--
 J. Hellenthal
The fact that there's a highway to Hell but only a stairway to Heaven says a lot about anticipated traffic volume. 


On Dec 10, 2020, at 12:41, Mel Beckman <mel () beckman org> wrote:


Jeez... some guys seem to take a joke literally - while ignoring a real and present danger - which was the point. 

Miles,
With all due respect, you didn’t present this as a joke. You presented "AI self-healing systems gone wild” as a genuine risk. Which it isn’t. In fact, AI fear mongering is a seriously debilitating factor in technology policy, where policymakers and pundits — who also don’t get “the joke” — lobby for silly laws and make ridiculous predictions, such as Elon Musks claim that, by 2025, “AI will be where AI conscious and vastly smarter than humans.”
That’s the kind of ignorance that will waste billions of dollars. No joke. 

 -mel
On Dec 10, 2020, at 8:47 AM, Miles Fidelman <mfidelman () meetinghouse net <mailto:mfidelman () meetinghouse net>> wrote:
Ahh.... invasive spambots, running on OpenStack ... "the telephone bell is tolling... " 

Miles

adamv0025@netconsultings.comwrote:
>Automated resource discovery + automated resource allocation = recipe for disaster 
That is literally how OpenStack works.
For now, don’t worry about AI taking away your freedom on its own, rather worry about how people using it might… 
adam
*From:*NANOG<nanog-bounces+adamv0025=netconsultings.com () nanog org>*On Behalf Of*Miles Fidelman 
*Sent:*Thursday, December 10, 2020 2:44 PM
*To:*'NANOG'<nanog () nanog org>
*Subject:*Re: The Real AI Threat?
adamv0025 () netconsultings com <mailto:adamv0025 () netconsultings com>wrote: 

    > Put them together, and the nightmare scenario is:

    > - machine learning algorithm detects need for more resources

    All good so far
> - machine learning algorithm makes use of vulnerability analysis library 
    > to find other systems with resources to spare, and starts attaching

    > those resources

    Right so a company would built, trained and fine-tuned an AI,
    or would have bought such a product and implemented it as part
    of its NMS/DDoS mitigation suite, to do the above?
    What is the probability of anyone thinking that to be a good idea?
    To me that does sound like an AI based virus rather than a tool
    one would want to develop or buy from a third party and then
    integrate into the day to day operations.
    You can’t take for instance alpha-0 or GPT-3 and make it do the
    above. You’d have to train it to do so over millions of
    examples and trials.
    Oh and also these won’t “wake up” one day and “think” to
    themselves oh I’m fed up with Atari games I’m going to learn
    myself some chess and then do some reading on wiki about the
    chess rules.
Jeez... some guys seem to take a joke literally - while ignoring a real and present danger - which was the point.
Meanwhile, yes, I think that a poorly ENGINEERED DDoS mitigation suite might well have failure modes that just keep eating up resources until systems start crashing all over the place.  Heck, spinning off processes until all available resources have been exhausted has been a failure mode of systems for years.  Automated resource discovery + automated resource allocation = recipe for disaster.  (No need for AIs eating the world.) 

Miles





--
In theory, there is no difference between theory and practice.
In practice, there is.  .... Yogi Berra
Theory is when you know everything but nothing works.
Practice is when everything works but no one knows why.
In our lab, theory and practice are combined:
nothing works and no one knows why.  ... unknown



--
In theory, there is no difference between theory and practice.
In practice, there is.  .... Yogi Berra

Theory is when you know everything but nothing works.
Practice is when everything works but no one knows why.
In our lab, theory and practice are combined:
nothing works and no one knows why.  ... unknown





--
In theory, there is no difference between theory and practice.
In practice, there is.  .... Yogi Berra

Theory is when you know everything but nothing works.
Practice is when everything works but no one knows why.
In our lab, theory and practice are combined:
nothing works and no one knows why.  ... unknown
Previous By Date Next
Previous By Thread Next

Current thread: