FSFE/fsfe-website
30
33
Fork 96
Code Issues 65 Pull Requests 14 Wiki Activity

Add article about Artificial Intelligence
continuous-integration/drone/push Build is passing
Details
continuous-integration/drone/pr Build is passing
Details

Browse Source
This commit is contained in:
vincent
2021-04-15 21:29:37 +02:00
parent e19ff089db
commit 9951b54efd
1 changed files with 324 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
freesoftware/artificial-intelligence.en.xhtml
+324
View File
@@ -0,0 +1,324 @@
<?xml version="1.0" encoding="UTF-8" ?>
<html>
<version>1</version>
<head>
<title>Artificial Intelligence and Free Software</title>
</head>
<body>
<h1>
Controlling technology at the age of Artificial Intelligence: a Free
Software perspective
</h1>
<div id="introduction">
<p>
Technical improvements, the accumulation of large, detailed
datasets and advancement in computer hardware have
lead to an Artificial Intelligence (AI) revolution.
For example, breakthroughs in computer vision have
enabled automated decision making based on images
and videos, the building of large datasets and
amelioration in text analysis coupled with the
gathering of personal data have given birth to countless AI
applications. These new AI applications have given lot of
benefits to These new AI applications have given lot of
benefits to European Union (EU) citizens. However, because
of its inherent complexity and requirements in technical
resources and knowledge, AI may undermine our ability to
control technology and put fundamental freedoms at risk.
Therefore, introducing new legislation on AI is a worthwhile
objective.
</p>
<p>
In the context of a new legislation, this article explains
how releasing AI applications under Free Software licenses pave
the way for more accessibility, transparency and fairness.
</p>
</div>
<h2 id="freesoftware">What is Free Software?</h2>
<p>
<a href="freesoftware.html">Free Software</a> (also known as Open Source) empowers people to control
technology by granting four freedoms to each user:
</p>
<ol>
<li>
The freedom to use software for any purpose, without
geographical limitations;
</li>
<li>
The freedom to study software, without any non-disclosure
agreement;
</li>
<li>The freedom to share software and copy it at no cost;</li>
<li>The freedom to improve software and share the improvements.</li>
</ol>
<p>
These freedoms are granted by releasing software under a Free
Software license, whose terms are compatible with the aforementioned
freedoms. There exists multiple Free Software licenses with
different goals. A software may be licensed under more than one
license. Because in order to be freely modified, an AI requires its
training code and the data, both needs to be released under a Free
Software license to consider the AI as being Free.
</p>
<h2 id="accessibility">Accessibility</h2>
<p>
Accessibility for AI means making it reusable, so that everyone may
tinker with it, improve it and use for their own means. To make AI
reusable, it can be released under a Free Software license. The
advantages of this approach are plenty. By having open legal
grounds, a Free AI fosters innovation, because one does not have to
deal with artificial restrictions that prevent people from reusing
work. Making AI Free therefore saves everyone from having to
reinvent the wheel, making researchers and developers alike able to
focus on creating new, better AI software instead of rebuilding
blocks and reproducing previous work again and again. In addition to
improving efficiency, by sharing expertise, Free AI also lowers the
cost of development by saving time and removing license fees. All of
this improves accessibility of AI, which leads to better and more
democratic solutions as everyone can participate.
</p>
<p>
Making AI reusable also makes it easier to build specialized AI
model upon more generic ones. If a generic AI model is released as
Free Software, rather than training a new model from scratch, one
could leverage the generic model as a starting point for a specific,
downstream prediction task. For example, one could use a generic
computer vision model [<a href="#ref-he_deep_2015">1</a>, <a href="#ref-simonyan_very_2015">2</a>]
as a starting point for managing public infrastructure which requires
specific image treatments. Just like with accessibility in general, this
approach has a key advantage: generic models with a lot of parameters
and trained on large datasets may make the downstream task easier to
learn. This makes AI more accessible by lowering the barrier to entry by
making it easier to reuse works.
</p>
<p>
However, making both the source code used to train the AI and the
corresponding data Free is sometimes not enough to make it accessible.
AI requires a huge amount of data in order to identify patterns and
correlations which lead to correct predictions. In contrary, not having
enough data reduces its ability to understand the world. Furthermore,
big datasets and their inherent complexity tend to make AI models large,
making their training time-consuming and resources intensive. The
complexity in handling the data required to train AI models, coupled
with the knowledge required to develop them and manage computer
resources demand a lot of human resources. Therefore, it may be hard to
exercise the freedoms offered by a Free AI, even though its training
source code and data might be released as Free Software. In those cases,
releasing the trained AI models as Free Software would greatly improve
accessibility.
</p>
<p>
Finally, it should be noted that, just like any other technology, making
AI reusable by everyone can potentially be harmful. For example, reusing
a face detector released as Free Software as part of a facial
recognition software can cause human right issues. However, this holds
true regardless of the technology involved. If a software use case is
deemed harmful, it should therefore be prohibited without an explicit
ban on AI technology.
</p>
<h2 id="transparency">Transparency</h2>
<p>
AI transparency can be subdivided in openness and interpretability.
In this context, openness is defined as the right to be
informed about the AI software, and interpretability is
defined as being able to understand how the input is
processed so that one can identify the factors taken into
account to make predictions, and their relative importance. In
Europe, the right to be informed about the decision of an algorithm
is granted by the Recital 71 of the General Data Protection
Regulation (GDPR) 2016/679 “
<em>
In any case, such processing should be subject to suitable
safeguards, which should include specific information to
the data subject and the right to obtain human
intervention, to express his or her point of view, to
obtain an explanation of the decision reached after such
assessment and to challenge the decision
</em>
.” Transparency can thus be defined as the ability to understand
what led to the predictions.
</p>
<p>
AI needs to be transparent because it is used for critical matters. For
example, it is used to determine credit worthiness
<a href="#ref-dastile_statistical_2020">[3]</a>, in self-driving cars
<a href="#ref-badue_self-driving_2019">[4]</a>, in predictive policing
<a href="#ref-ensign_runaway_2018">[5]</a> or in healthcare
<a href="#ref-schwalbe_artificial_2020">[6]</a>. In these contexts, getting information about how the predictions are done
is therefore critical and information about the data used and how it was
processed by the AI should be made available. Moreover, trust and adoption
of AI would consequently be higher. Furthermore, modern AI technologies such
as deep learning are not meant to be transparent, because are composed of
millions or billions of individual parameters <a href="#ref-canziani_analysis_2017">[7]</a>, making them very complex and hard to understand. This calls for Free
Software which seeks to analyze this complexity.
</p>
<p>
Technologies released as Free Software to make AI more transparent already
exists. For example, Local Interpretable Model-Agnostic Explanations (LIME)
<a href="#ref-ribeiro_why_2016">[8]</a>
is a software package which simplifies a complex prediction model by
simulating it with a simpler, more interpretable version, thus enabling
users of the AI to understand the parameters that played a role in the
prediction. Captum <a href="#ref-kokhlikyan_captum_2020">[9]</a>
is library released as Free Software providing an attribution mechanism
allowing one to understand the relative importance of each input variable
and each parameter of a deep learning model. Making AI more transparent is
therefore possible.
</p>
<p>
Although a proprietary AI can be transparent, Free Software facilitates this
process by making auditing and inspection easier. While some data might be
too sensitive to be released under a Free Software license, statistical
properties of the data can still be published. With Free Software, everyone
is able to run the AI to understand how it is made, and look up the data
that went through it. However, it should be noted that the AI model itself,
being composed of millions or billions of parameters, is not meant to be
transparent. But simulating the AI model with a much simpler one would make
it easy to inspect it.
</p>
<p>
Another benefit of Free Software in this context is that by granting the
right to improve the AI software and share improvements with others, it
allows everybody to improve transparency, thereby preventing vendor lock-in
where one has to wait until the software provider makes AI more transparent.
</p>
<h2 id="fairness">Fairness</h2>
<p>
In artificial intelligence (AI), fairness is defined as making it free of
harmful discrimination based on one’s sensitive characteristics such as
gender, ethnicity, religion, disabilities or sexual orientation. Because AI
models are trained on datasets containing human behaviors and activities
that can be unfair, and AI models are designed to recognize and reproduce
existing patterns, they can create harmful discrimination and human right
violations. For example, (COMPAS) <a href="#ref-noauthor_practitioners_2015">[10]</a>, an algorithm attributing scores which indicates how likely one is going to
recidivate their crime, was found to be unfair towards African American
<a href="#ref-mattu_machine_2015">[11]</a>, because for them, 44.9% of cases were false positives. The algorithm
attributed a high change of recidivism despite the defendants not
re-offending. Conversely, 47.7% of the cases for white people were labeled
as low risk of recidivism despite them re-offending. Suspected unfairness
has also been found in healthcare
<a href="#ref-obermeyer_dissecting_2019">[12]</a>, where an algorithm was used to attribute risks scores to patients, thereby
identifying those needing additional care resources. To have the same risks
scores as white people, black people needed to be in an worst health
situation, in term of severity in hypertension, diabetes, anemia, bad
cholesterol, or renal failure. Therefore, real fairness issues exist in AI
algorithm. Moreover, from a legal perspective, checking for fairness issues
is required by the Recital 71 of the GDPR, which requires to “
<em>
prevent, inter alia, discriminatory effects on natural persons on the basis
of racial or ethnic origin, political opinion, religion or beliefs, trade
union membership, genetic or health status or sexual orientation, or
processing that results in measures having such an effect.
</em>
” We thus need solutions to detect potential fairness issues in datasets on
which AI is trained and correct it when it occurs.
</p>
<p>
To detect fairness, one needs to quantify it. There are lots of ways to
define fairness for AI, based on two categories of approaches. The first one
verifies that people grouped according to some sensitive characteristic are
treated similarly by the algorithm, e.g. in term of accuracy, true positive
rate and false positive rate. The second approach measures fairness at the
individual level by ensuring that similar individuals are treated similarly
by the algorithm <a href="#ref-dwork_fairness_2011">[13]</a>. More formally, a distance measure between samples of the dataset and a
distance measure between the predictions of the algorithm are compared to
ensure their ratio is consistent. However, satisfying group fairness and
individual fairness at the same time might be impossible <a href="#ref-kleinberg_inherent_2016">[14]</a>. There are three commonly used methods to mitigate unfairness, if detected:
</p>
<ol>
<li>
Remove the sensitive attribute (e.g. gender, ethnicity, religion, etc.)
from the dataset. This approach does not work in real-world scenario
because removing the sensitive attribute is not enough to completely
mask it, as the sensitive attribute is often correlated with other
attributes of the dataset. Removing it is therefore not sufficient, and
removing all attributes correlated with it leads to a lot of information
loss;
</li>
<li>
Ensure that the dataset has an equal representation of people if grouped
by a sensitive characteristic;
</li>
<li>
Optimize the AI model for accuracy and fairness at the same time. While
the algorithm is trained on an existing dataset that contains unfair
discrimination, both consider its accuracy and its fairness
<a href="#ref-zafar_fairness_2017">[15]</a>. In other words, add fairness to the goal of the algorithm.
</li>
</ol>
<p>
If those methods are used, having a perfectly accurate and fair algorithm is
impossible, but if the accuracy is defined on a dataset that is known to
contain unfair treatment of a particular group, having a less than perfect
accuracy may be deemed acceptable.
</p>
<p>
Because a AI released as Free Software may be used and inspected by
everyone, verify if it is free of potentially harmful discrimination is
easier than if it were proprietary. Moreover, this synergies with AI
transparency (see Section <a href="#transparency">Transparency</a>), as a transparent AI facilitates the understanding of the factors considered
for making predictions. While necessary, releasing AI as Free Software does not
make fair, but make fairness easier to evaluate and enforce.
</p>
<h2 id="conclusions">Conclusions</h2>
<p>
In this article, we highlighted potential issues around the democratization
of artificial intelligence (AI) and implications for human rights. Possible
Free Software solutions are presented to tackle these issues. In particular,
we showed that AI needs to be accessible, transparent and fair in order to
be usable. While not a sufficient solution, releasing AI under Free Software
licenses is necessary for its widespread use throughout our information
systems by making it more scrutable, trustworthy and safe for everyone.
</p>
<h2 id="references">References</h2>
<div id="ref-he_deep_2015">
[1] K. He, X. Zhang, S. Ren, and J. Sun, “Deep Residual Learning for Image Recognition,” <em>arXiv:1512.03385 [cs]</em>, Dec. 2015.</div>
<div id="ref-simonyan_very_2015">
[2] K. Simonyan and A. Zisserman, “Very Deep Convolutional Networks for Large-Scale Image Recognition,” <em>arXiv:1409.1556 [cs]</em>, Apr. 2015.</div>
<div id="ref-dastile_statistical_2020">
[3] X. Dastile, T. Celik, and M. Potsane, “Statistical and machine learning models in credit scoring: A systematic literature survey,” <em>Applied Soft Computing</em>, vol. 91, p. 106263, 2020, doi: <a href="https://doi.org/10.1016/j.asoc.2020.106263">10.1016/j.asoc.2020.106263</a>.</div>
<div id="ref-badue_self-driving_2019">
[4] C. Badue <em>et al.</em>, “Self-Driving Cars: A Survey,” <em>arXiv:1901.04407 [cs]</em>, Oct. 2019.</div>
<div id="ref-ensign_runaway_2018">
[5] D. Ensign, S. A. Friedler, S. Neville, C. Scheidegger, and S. Venkatasubramanian, “Runaway Feedback Loops in Predictive Policing,” in <em>Conference on Fairness, Accountability and Transparency</em>, Jan. 2018, pp. 160–171.</div>
<div id="ref-schwalbe_artificial_2020">
[6] N. Schwalbe and B. Wahl, “Artificial intelligence and the future of global health,” <em>The Lancet</em>, vol. 395, no. 10236, pp. 1579–1586, May 2020, doi: <a href="https://doi.org/10.1016/S0140-6736(20)30226-9">10.1016/S0140-6736(20)30226-9</a>.</div>
<div id="ref-canziani_analysis_2017">
[7] A. Canziani, A. Paszke, and E. Culurciello, “An Analysis of Deep Neural Network Models for Practical Applications,” <em>arXiv:1605.07678 [cs]</em>, Apr. 2017.</div>
<div id="ref-ribeiro_why_2016">
[8] M. T. Ribeiro, S. Singh, and C. Guestrin, “"Why Should I Trust You?": Explaining the Predictions of Any Classifier,” <em>arXiv:1602.04938 [cs, stat]</em>, Aug. 2016.</div>
<div id="ref-kokhlikyan_captum_2020">
[9] N. Kokhlikyan <em>et al.</em>, <em>Captum: A unified and generic model interpretability library for PyTorch</em>. 2020.</div>
<div id="ref-noauthor_practitioners_2015">
[10] “Practitioners Guide to COMPAS.” Northpointe, Mar. 2015.</div>
<div id="ref-mattu_machine_2015">
[11] L. K. Mattu Jeff Larson, “Machine Bias,” <em>ProPublica</em>. Mar. 2015.</div>
<div id="ref-obermeyer_dissecting_2019">
[12] Z. Obermeyer, B. Powers, C. Vogeli, and S. Mullainathan, “Dissecting racial bias in an algorithm used to manage the health of populations,” <em>Science (New York, N.Y.)</em>, vol. 366, no. 6464, pp. 447–453, Oct. 2019, doi: <a href="https://doi.org/10.1126/science.aax2342">10.1126/science.aax2342</a>.</div>
<div id="ref-dwork_fairness_2011">
[13] C. Dwork, M. Hardt, T. Pitassi, O. Reingold, and R. Zemel, “Fairness Through Awareness,” <em>arXiv:1104.3913 [cs]</em>, Nov. 2011.</div>
<div id="ref-kleinberg_inherent_2016">
[14] J. Kleinberg, S. Mullainathan, and M. Raghavan, “Inherent Trade-Offs in the Fair Determination of Risk Scores,” <em>arXiv:1609.05807 [cs, stat]</em>, Nov. 2016.</div>
<div id="ref-zafar_fairness_2017">
[15] M. B. Zafar, I. Valera, M. G. Rodriguez, and K. P. Gummadi, “Fairness Beyond Disparate Treatment &amp; Disparate Impact: Learning Classification without Disparate Mistreatment,” <em>Proceedings of the 26th International Conference on World Wide Web</em>, pp. 1171–1180, Apr. 2017, doi: <a href="https://doi.org/10.1145/3038912.3052660">10.1145/3038912.3052660</a>.</div>
</body>
</html>