What bleak topic to engage with... Today's uplifting news is that all
the machine learned intelligence needed to roll all this out on the
large scale does rely on very complex algorithms which have severe
issues with being too dependent on the initial condition. While the
paper linked below describes the findings as a means to improve the
learning algorithms it points to a vector to hack the AI. At the moment
a small yellow, pink, or green pixels wins :)
https://arxiv.org/abs/1710.08864
One pixel attack for fooling deep neural networks
Jiawei Su, Danilo Vasconcellos Vargas, Sakurai Kouichi (Submitted on 24
Oct 2017)
/Recent research has revealed that the output of Deep neural
networks(DNN) is not continuous and very sensitive to tiny perturbation
on the input vectors and accordingly several methods have been proposed
for crafting effective perturbation against the networks. In this paper,
we propose a novel method for optically calculating extremely small
adversarial perturbation (few-pixels attack), based on differential
evolution. It requires much less adversarial information and works with
a broader classes of DNN models. The results show that 73.8% of the test
images can be crafted to adversarial images with modification just on
one pixel with 98.7% confidence on average. In addition, it is known
that investigating the robustness problem of DNN can bring critical
clues for understanding the geometrical features of the DNN decision map
in high dimensional input space. The results of conducting few-pixels
attack contribute quantitative measurements and analysis to the
geometrical understanding from a different perspective compared to
previous works./
On 30/10/2017 21:47, Morlock Elloi wrote:
To throw in two items, one presently real, and one somewhat
speculative. Both are contingent on high speed network-to-brain (N2B)
interface, namely a handset, which has victim's attention many hours
every day.
1. Social networks (ie. FB) likely know your IQ with margin of error
of 5 points or less.
IQ is hard to mask, unnoticeable tests can be easily implemented,
probably focusing on the speed of actions, ie. figuring out where the
button is in a slightly changed interface, etc., which can be done
over long time, not in one sitting. This information did not exist
before (national IQ dataset), has nothing to do with your habits, and
is highly valuable: once FB separates sharpies from dims (exactly half
of us are below average), it can use different strategies to influence
each. More importantly, this data is valuable to the law enforcement:
if you are looking to frame someone, you go for dims. If you are
looking for leaders, you narrow your attention to sharpies.
2. Ubiquitous N2B interfaces may enable effective brain hacking.
We are not talking advertizing and nudging here, but straightforward
hacking that bypasses voluntary/consciousness layers. After all, the
brain is just a computer, and it's a matter of time before buffer
overflow zero days are figured out (note that they will stay zero
days, as there is no one to send you the patch.) To illustrate the
principle, this could be similar to the way that flashing patterns
induce epileptic attacks in those prone to them. I don't expect a good
brain overflow hack to have crude flashing patterns, it may have
something far more discreet, a combination of outputs and feedbacks
(something comes up on the screen, you click on X, then something else
comes up, you ... etc.) that causes ... something. I'm pretty sure
that self-respecting TLAs are already investing billions in the
research (they did spend $90M on LSD research in 1950s.) The presence
of the N2B interface is just too important to ignore.
But if the people in power are using these algorithms to quietly watch
us, to judge us and to nudge us, to predict and identify the
troublemakers and the rebels, to deploy persuasion architectures at
scale and to manipulate individuals one by one using their personal,
individual weaknesses and vulnerabilities, and if they're doing it at
scale through our private screens so that we don't even know what our
fellow citizens and neighbors are seeing, that authoritarianism will
envelop us like a spider's web and we may not even know we're in it.
# distributed via <nettime>: no commercial use without permission
# <nettime> is a moderated mailing list for net criticism,
# collaborative text filtering and cultural politics of the nets
# more info: http://mx.kein.org/mailman/listinfo/nettime-l
# archive: http://www.nettime.org contact: [email protected]
# @nettime_bot tweets mail w/ sender unless #ANON is in Subject:
--
DE: +49 (0)160 9549 5269
UK: +44 (0)75 0655 0520
http://vincentvanuffelen.com
http://transmit-interfere.com
http://deepmediaresearch.org
# distributed via <nettime>: no commercial use without permission
# <nettime> is a moderated mailing list for net criticism,
# collaborative text filtering and cultural politics of the nets
# more info: http://mx.kein.org/mailman/listinfo/nettime-l
# archive: http://www.nettime.org contact: [email protected]
# @nettime_bot tweets mail w/ sender unless #ANON is in Subject:
