Human rights in the age of

IMAGE
LABELS
HIDDEN LAYERS
OUTPUT:
“ELEPHANT”
INPUT
FIG 1. The ‘neural networks’ structure

---

accessnow.org
10
HUMAN RIGHTS IN THE AGE OF ARTIFICIAL INTELLIGENCE
Progress in deep learning has been the cause for much of the optimism about AI due its ability to process 
and find patterns in massive amounts of data with accuracy.
12
Whereas early ML typically uses a decision-
tree structure, deep learning has become the dominant technique. It is often used to power specific ML 
approaches such as machine vision and natural language processing.
13
5.1. MACHINE VISION: A specific ML approach that allows computers to recognize and evaluate images.
14
It 
is used by Google to help you search images and by Facebook to automatically tag people in photos. 
5.2. NATURAL LANGUAGE PROCESSING: A specific ML approach that helps computers understand, 
interpret, and manipulate human language. It does this by breaking down language into shorter pieces and 
discovering how the pieces fit together to create meaning. Natural language processing enables commonly 
used services such as Google Translate and chatbots.
15
5.3 SPEECH RECOGNITION:  A specific ML approach allows computers to translate spoken language into 
text.
16
It allows you to use talk-to-text on your smartphone. It is often paired together with natural 
language processing and is used to power virtual assistants like Siri and Alexa.
6. ALGORITHM: 
At its most simple, an algorithm is “a set of guidelines that describe how to perform a 
task.”
17
Within computer science, an algorithm is a sequence of instructions that tell a computer what to 
do.
18
AI works through algorithms (neural networks are a type of algorithm), but not all algorithms involve 
artificial intelligence. 
12 Preparing for the Future of Artificial Intelligence”, 9–10.
13 Virtual assistants Siri, Cortana, Alexa use neural networks for speech recognition and to imitate human conversation. Deep learning in this case allows 
these virtual assistants to detect and understand the nuances of speech and produce a response that feels conversational. See https://machinelearning.apple.
com/2017/08/06/siri-voices.html for more information. IBM’s Watson uses deep learning techniques for machine vision to analyze to quickly and accurately 
interpret medical information and help doctors diagnose disease. See https://www.ibm.com/watson/health/ for more information.
14 “What Is a Machine Vision System (MVS)? - Definition from Techopedia,” Techopedia.com, accessed May 12, 2018, https://www.techopedia.com/
definition/30414/machine-vision-system-mvs.
15 “ “What Is Natural Language Processing?,” accessed May 12, 2018, https://www.sas.com/en_us/insights/analytics/what-is-natural-language-
processing-nlp.html.
16 “Speech Recognition,” Wikipedia, May 1, 2018, https://en.wikipedia.org/w/index.php?title=Speech_recognition&oldid=839191878.
17 “What Is an Algorithm? An Explainer.,” accessed May 12, 2018, http://www.slate.com/articles/technology/future_tense/2016/02/what_is_an_
algorithm_an_explainer.html.
18 Ibid.
FIG 2. The relationships of the concepts in artificial intelligence

---

accessnow.org
11