Clinton Foundation 42
Clubcard (Tesco) 26
Cohen’s Kappa 215n12
cold cases 172
Cold War 18
Colgan, Steyve 155
Commodore 64 ix
COMPAS algorithm 63, 64
ProPublica analysis
accuracy of scores 65
false positives 66
mistakes 65–8
racial groups 65–6
secrecy of 69
CompStat 149
computational statistics 12
computer code 8
computer intelligence 13 see also AI (artificial intelligence)
computer science 8
computing power 5
considered thought 72
cookies 34
Cope, David 189, 190–1, 193
cops on the dots 155–6
Corelogic 31
counter-intuition 122
creativity, human 192–3
Creemers, Rogier 46
creepy line 28, 30, 39
crime 141–73
algorithmic regulation 173
boost effect 151, 152
burglary 150–1
cops on the dots 155–6
geographical patterns 142–3
gun 158
hotspots 148, 149, 150–1, 155
HunchLab algorithm 157–8
New York City subway 147–50
predictability of 144
PredPol algorithm 152–7, 158
proximity of offenders’ homes 144
recognizable patterns 143–4
retail 170
Strategic Subject List 158
target hardening 154–5 see also facial recognition
crime data 143–4
Crimewatch programme 142
criminals
buffer zone 144
distance decay 144
knowledge of local geographic area 144
serial offenders 144, 145
customers
data profiles 32
inferred data 32–4
insurance data 30–1
shopping habits 28, 29, 31
supermarket data 26–8
superstore data 28–31
cyclists 129
Daimler 115, 130
DARPA (US Defence Advanced Research Projects Agency)
driverless cars 113–16
investment in 113
Grand Challenge (2004) 113–14, 117
course 114
diversity of vehicles 114
GPS coordinates 114
problems 114–15
top-scoring vehicle 115
vehicles’ failure to finish 115
Grand Challenge (2005) 115
targeting of military vehicles 113–14
data 25–47
exchange of 25, 26, 44–5
dangers of 45
healthcare 105
insurance 30–1
internet browsing history 36–7, 36–8
internet giants 36
manipulation and 39–44
medical records 102–7
benefits of algorithms 106
DeepMind 104–5
disconnected 102–3
misuse of data 106
privacy 105–7
patterns in 79–81, 108
personal 108
regulation of
America 46–7
Europe 46–7
global trend 47
sale of 36–7
Sesame Credit 45–6, 168
shopping habits 28, 29, 31
supermarkets and 26–8
superstores and 28–31
data brokers 31–9
benefits provided by 32
Cambridge Analytica 39–42
data profiles 32
inferred data 32–4, 35
murky practices of 47
online adverts 33–5
rich and detailed datasets 103
Sesame Credit 45–6
unregulated 36
in America 36
dating algorithms 9
Davies, Toby 156, 157
decision trees 56–8
Deep Blue 5–7, 8
deep learning 86
DeepMind
access to full medical histories 104–5
consent ignored 105
outrage 104
contract with Royal Free NHS Trust 104
dementia 90–2
Dewes, Andreas 36–7
Dhami, Mandeep 75, 76
diabetic retinopathy 96
Diaconis, Pesri 124
diagnostic machines 98–101, 110–11
differential diagnosis 99
discrimination 71
disease
Alzheimer’s disease 90–1, 92
diagnosing 59, 99, 100
hereditary causes 108
Hippocrates’s understanding of 80
Huntington’s disease 110
motor neurone disease 100
pre-modern medicine 80 see also breast cancer
distance decay 144
DNA (deoxyribonucleic acid) 106, 109
testing 164–5
doctors 81
unique skills of 81–2
Dodds, Peter 176–7
doppelgängers 161–3, 164, 169
Douglas, Neil 162–3
driver-assistance technology 131
driverless cars 113–40
advantages 137
algorithms and 117
Bayes’ red ball analogy 123–4
ALVINN (Autonomous Land Vehicle In a Neural Network) 118–19
autonomy 129, 130
full 127, 130, 134, 138
Bayes’ theorem 121–4
breaking the rules of the road 128
bullying by people 129
cameras and 117–18
conditions for 129
cyclists and 129
dealing with people 128–9
difficulties of building 117–18, 127–8
early technology 116–17
framing of technology 138
inevitability of errors 140
measurement 119, 120
neural networks 117–18
potential issues 116
pre-decided go-zones 130
sci-fi era 116
simulations 136–7
speed and direction 117
support for drivers 139
trolley problem 125–6
Uber 135
Waymo 129–30
driverless technology 131
Dubois, Captain 133, 137
Duggan, Mark 49
Dunn, Edwina 26
early warning systems 18
earthquakes 151–2
eBureau 31
Eckert, Svea 36–7
empathy 81–2
ensembles 58
Eppink, Richard 17, 18
Epstein, Robert 14–15
equations 8
Equivant (formerly Northpointe) 69, 217n38
errors in algorithms 18–19, 61–2, 76, 159–60, 197–9, 201
false negatives 62, 87, 88
false positives 62, 66, 87, 88
Eureka Prometheus Project 117
expectant mothers 28–9
expectations 7
Experiments in Musical Intelligence (EMI) 189–91, 193
Face ID (Apple) 165–6
Facebook 2, 9, 36, 40
filtering 10
Likes 39–40
news feeds experiment 42–3
personality scores 39
privacy issues 25
severing ties with data brokers 47
FaceFirst 170, 171
FaceNet (Google) 167, 169
facial recognition
accuracy 171
falling 168
increasing 169
algorithms 160–3, 165, 201–2
2D images 166–7
3D model of face 165–6
&
nbsp; Face ID (Apple) 165–6
FaceFirst 170
FaceNet (Google) 167, 169
measurements 163
MegaFace 168–9
statistical approach 166–7
Tencent YouTu Lab 169
in China 168
cold cases 172
David Baril incident 171–2
differences from DNA testing 164–5
doppelgängers 161–3, 164, 169
gambling addicts 169–70
identical looks 162–3, 164, 165
misidentification 168
neural networks 166–7
NYPD statistics 172
passport officers 161, 164
police databases of facial images 168
resemblance 164, 165
shoplifters 170
pros and cons of technology 170–1
software 160
trade-off 171–3
Youssef Zaghba incident 172
fairness 66–8, 201
tweaking 70
fake news 42
false negatives 62, 87, 88
false positives 62, 66, 87, 88
FBI (Federal Bureau of Investigation) 168
Federal Communications Commission (FCC) 36
Federal Trade Commission 47
feedback loops 156–7
films 180–4
algorithms for 183
edits 182–3
IMDb website 181–2
investment in 180
John Carter (film) 180
novelty and 182
popularity 183–4
predicting success 180–1
Rotten Tomatoes website 181
study 181–2
keywords 181–2
filtering algorithms 9–10
Financial Times 116
fingerprinting 145, 171
Firebird II 116
Firefox 47
Foothill 156
Ford 115, 130
forecasts, decision trees 57–8
free technology 44
Fuchs, Thomas 101
Galton, Francis 107–8
gambling addicts 169–70
GDPR (General Data Protection Regulation) 46
General Motors 116
genetic algorithms 191–2
genetic testing 108, 110
genome, human 108, 110
geographical patterns 142–3
geoprofiling 147
algorithm 144
Germany
facial recognition algorithms 161
linking of healthcare records 103
Goldman, William 181, 184
Google 14–15, 36
creepy line 28, 30, 39
data security record 105
FaceNet algorithm 167, 169
high-paying executive jobs 35 see also DeepMind
Google Brain 96
Google Chrome plugins 36–7
Google Images 69
Google Maps 120
Google Search 8
Google Translate 38
GPS 3, 13–14, 114
potential errors 120
guardian mode 139
Guerry, André-Michel 143–4
gun crime 158
Hamm, John 99
Hammond, Philip 115
Harkness, Timandra 105–6
Harvard researchers experiment (2013) 88–9
healthcare
common goal 111–12
exhibition (1884) 107
linking of medical records 102–3
sparse and disconnected dataset 103
healthcare data 105
Hinton, Geoffrey 86
Hippocrates 80
Hofstadter, Douglas 189–90, 194
home cooks 30–1
homosexuality 22
hotspots, crime 148, 149, 150–1, 155
Hugo, Christoph von 124–5
human characteristics, study of 107
human genome 108, 110
human intuition 71–4, 77, 122
humans
and algorithms
opposite skills to 139
prediction 22, 59–61, 62–5
struggle between 20–4
understanding the human mind 6
domination by machines 5–6
vs machines 59–61, 62–4
power of veto 19
PredPol (PREDictive POLicing) 153–4
strengths of 139
weaknesses of 139
Humby, Clive 26, 27, 28
Hume, David 184–5
HunchLab 157–8
Huntington’s disease 110
IBM 97–8 see also Deep Blue
Ibrahim, Rahinah 197–8
Idaho Department of Health and Welfare
budget tool 16
arbitrary numbers 16–17
bugs and errors 17
Excel spreadsheet 17
legally unconstitutional 17
naive trust 17–18
random results 17
cuts to Medicaid assistance 16–17
Medicaid team 17
secrecy of software 17
Illinois prisons 55, 56
image recognition 11, 84–7, 211n13
inferred data 32–4, 35
personality traits 40
Innocence Project 164
Instagram 36
insurance 30–1
genetic tests for Huntington’s disease 110
life insurance stipulations 109
unavailability for obese patients 106
intelligence tracking prevention 47
internet browsing history 36–8
anonymous 36, 37
de-anonymizing 37–8
personal identifiers 37–8
sale of 36–7
Internet Movie Database (IMDb) 181–2
intuition see human intuition
jay-walking 129
Jemaah Islam 198
Jemaah Islamiyah 198
Jennings, Ken 97–8
Jeopardy (TV show) 97–9
John Carter (film) 180
Johnson, Richard 50, 51
Jones Beach 1
Jones, Robert 13–14
judges
anchoring effect 73
bail, factors for consideration 73
decision-making
consistency in 51
contradictions in 52–3
differences in 52
discretion in 53
unbiased 77
discrimination and bias 70–1, 75
intuition and considered thought 72
lawyers’ preference over algorithms 76–7
vs machines 59–61
offenders’ preference over algorithms 76
perpetuation of bias 73
sentencing 53–4, 63
use of algorithms 63, 64
Weber’s Law 74–5
Jukebox 192
junk algorithms 200
Just Noticeable Difference 74
justice 49–78
algorithms and 54–6
justification for 77
appeals process 51
Brixton riots 49–51
by country
Australia 53
Canada 54
England 54
Ireland 54
Scotland 54
United States 53, 54
Wales 54
discretion of judges 53
discrimination 70–1
humans vs machines 59–61, 62–4
hypothetical cases (UK research) 52–3
defendants appearing twice 52–3
differences in judgement 52, 53
hypothetical cases (US research) 51–2
differences in judgements 52
differences in sentencing 52
inherent injustice 77
machine bias 65–71
maximum terms 54
purpose of 77–8
re-offending 54, 55
reasonable doubt 51
rehabilitation 55
&nb
sp; risk-assessment algorithms 56
sentencing
consistency in 51
mitigating factors in 53
substantial grounds 51
Kadoodle 15–16
Kahneman, Daniel 72
Kanevsky, Dr Jonathan 93, 95
kangaroos 128
Kant, Immanuel 185
Kasparov, Gary 5–7, 202
Kelly, Frank 87
Kerner, Winifred 188–9
Kernighan, Brian x
Killingbeck 145, 146
Larson, Steve 188–9
lasers 119–20
Leibniz, Gottfried 184
Leroi, Armand 186, 192–3
level 0 (driverless technology) 131
level 1 (driverless technology) 131
level 2 (driverless technology) 131, 136
careful attention 134–5
level 3 (driverless technology) 131
technical challenge 136
level 4 (driverless technology) 131
level 5 (driverless technology) 131
Li Yingyun 45
Lickel, Charles 97–8
LiDAR (Light Detection and Ranging) 119–20
life insurance 109
‘Lockdown’ (52Metro) 177
logic 8
logical instructions 8
London Bridge 172
London School of Economics (LSE) 129
Loomis, Eric 217n38
Los Angeles Police Department 152, 155
Lucas, Teghan 161–2, 163
machine-learning algorithms 10–11
neural networks 85–6
random forests 58–9
machines
art and 194
bias in 65–71
diagnostic 98–101, 110–11
domination of humans 5–6
vs humans 59–61, 62–4
paradoxical relationship with 22–3
recognising images 84–7
superior judgement of 16
symbolic dominance over humans 5–6
Magic Test 200
magical illusions 18
mammogram screenings 94, 96
manipulation 39–44
micro-manipulation 42–4
Maple, Jack 147–50
Marx, Gary 173
mastectomies 83, 84, 92, 94
maternity wards, deaths on 81
mathematical certainty 68
mathematical objects 8
McGrayne, Sharon Bertsch 122
mechanized weaving machines 2
Medicaid assistance 16–17
medical conditions, algorithms for 96–7
medical records 102–7
benefits of algorithms 106
DeepMind 104–5
disconnected 102–3
misuse of data 106
privacy 105–7
medicine 79–112
in ancient times 80
cancer diagnoses study 79–80
complexity of 103–4
diabetic retinopathy 96
diagnostic machines 98–101, 110–11
choosing between individuals and the population 111
in fifteenth-century China 81
Hippocrates and 80
magic and 80
medical records 102–6
neural networks 85–6, 95, 96, 219–20n11
in nineteenth-century Europe 81
pathology 79, 82–3
patterns in data 79–81
predicting dementia 90–2
scientific base 80 see also Watson (IBM computer)
Meehl, Paul 21–2
MegaFace challenge 168–9