-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathindex.html
250 lines (164 loc) · 25.1 KB
/
index.html
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
<!DOCTYPE html>
<html>
<head>
<meta charset="utf-8">
<meta http-equiv="X-UA-Compatible" content="IE=edge">
<title>Human v machine creativity</title>
<link rel="stylesheet" href="https://maxcdn.bootstrapcdn.com/bootstrap/3.3.7/css/bootstrap.min.css">
</head>
<body>
<div class="container">
<h1>Human v machine creativity</h1>
<p class="lead">This story's content was written by a human, but its narrative was chosen for you by a machine.</p>
<p><div id="btnShuffle" class="btn btn-primary">Shuffle</div></p>
<div id="movableContainer" class="hide col-sm-8 col-sm-offset-2">
<div class="moveable">
<h4>Rembrandt</h4>
<p> In 2016, more than three hundred years after the death of Dutch artist Rembrandt van Rijm, a new Rembrandt painting appeared on the scene. It was not a forgotten relic – discovered an attic (like the “Forgotten Caravaggio”) or a found in private homes after they survived Nazi plundering – but a completely new painting. A man in 17th-century apparel replete with neck ruff and over-sized black hat stares bemusedly out of the canvas.</p>
<p>The painting, part of an advertising campaign, was born out of a deep-learning algorithm that analysed Rembrandt's 346 paintings. According to the public relations machinery – in this case a slick organisation of real people, rather than a complicated computational process – that handled the launch of the “Next Rembrandt”, a software system was trained to understand the Dutch master's use of geometry, composition and painting materials. “A facial recognition algorithm identified and classified the most typical geometric patterns used by Rembrandt to paint human features,” they say. “It then used the learned principles to replicate the style and generate new facial features for the painting.”</p>
<p>Art historian Gary Schwartz said it was “a fascinating exercise in connoisseurship”. “While no one will claim that Rembrandt can be reduced to an algorithm, this technique offers an opportunity to test your own ideas about his paintings in concrete, visual form.”</p>
<p>While the developers and art historians have created a painting that looks like it was created by Rembrandt, they are all quick to say that only the artist himself could create a true Rembrandt. But is the origin of an artwork more important than the aesthetic value of the painting itself?</p>
</div>
<div class="moveable">
<h4>Folktales</h4>
<p>In 1928, a thin book – fewer than 200 pages – hit the shelves in what was then Leningrad, and stayed there for 30 years, the province of Russian literary specialists and anthropologists. It was only once it was translated into English 30 years later that <a href="http://web.mit.edu/allanmc/www/propp.pdf">The Morphology of the Folktale</a> rose to its current prominent position in literary studies. In it, Russian Vladimir Propp painstakingly characterised the elements that form a folktale.</p>
<p>Humans have been telling each other these stories around firesides for centuries, and Propp showed that they are highly structured in a system of conventions, signs and models. From the functions and attributes of the main characters to the patterns that these tales follow, there is a system – an algorithm, if you will – that governs this art form. If the final story was put together by a computer, would it still be considered an art form?</p>
</div>
<div class="moveable">
<h4>Algorithmic creativity anathema</h4>
<p>“People often object to machine creativity because we reserve creativity for ourselves,” says Veale. “The idea that creativity is algorithmic is an anathema to people.”</p>
</div>
<div class="moveable">
<h4>Creativity definition</h4>
<p><a href="http://www.dictionary.com/browse/creativity">Creativity</a> |ˌkriːeɪˈtɪvɪti|</p>
<p>Noun<br>
the state or quality of being creative.<br>
the ability to transcend traditional ideas, rules, patterns,relationships, or the like, and to create meaningful new ideas, forms,methods, interpretations, etc.; originality, progressiveness, or imagination: the need for creativity in modern industry; creativity in the performing arts.<br>
the process by which one utilizes creative ability: Extensive reading stimulated his creativity.</p>
<p>Origin of creativity<br>
1870-1875</p>
</div>
<div class="moveable">
<h4>Turing test</h4>
<p>The war had been won, and in 1950 Turing was working in Victoria University's mathematics department in Manchester. Turing knew that one day we would be talking about whether machines could think. In a paper called “Computing Machinery and Intelligence” he laid out what has become the ultimate test for machine intelligence. The Turing Test, based on what Turing calls “the imitation game”, pivots on how well a machine can imitate a human, and whether an observer can tell the difference between a human's answer and a machine's.</p>
<p>But academics say that this test is holding back the field of computational creativity.</p>
<p>Pease says that the Turing Test is discriminatory.</p>
<p>“I don't think it is good for the field of computational creativity: creative behaviour and human behaviour overlap, but they are not the same thing,” she says.</p>
<p>From systems found in the natural world to the outputs of computers, these only pass the Turing Test if they can mimic human outputs. “We want to know the context, who created it, before we give a judgment on whether it is creative,” she says.</p>
</div>
<div class="moveable">
<h4>Cognitive cooking</h4>
<p>Ultimately the chef overruled the culinary inclinations of IBM's Watson: Porcini mushrooms and bacon, wrapped in a meringue, was too adventurous for the journalists assembled at the Griffin restaurant in Johannesburg in 2014. We were the guinea pigs to test the supercomputer's new “cognitive cooking” application, an attempt to introduce artificial intelligence to “creative industries”.</p>
<p>Watson, named after IBM's first chief executive Thomas J Watson, had been programmed with tens of thousands of recipes to allow it to understand what people like, which foods can be paired, and how to compose dishes, with a soupçon of cultural sensitivity thrown in.</p>
<p>“What appears as a list of ingredients for a novel and flavourful meal is actually the result of a system that intelligently generates millions of ideas out of quintillions of possibilities, and then predicts which ones are the most surprising and pleasant,” IBM said at the time.</p>
<p>The dishes – which ultimately ended up being a Thai-Swiss asparagus, which was surprising but with a suspicious texture, and pine nut-encrusted lamb, which was quite delicious – comprised three elements: a main ingredient (such as asparagus or lamb), the orientation (such as vegetarian) and a sliding scale of adventurousness.</p>
<p>Thomas Hughes, owner of the restaurant and the human hands Watson used to prepare the meal, said: “It goes against the way we [chefs] are programmed with food. We're trained to train our palettes, and that was something to overcome [in following Watson's instructions].”</p>
<p>Shane Radford, IBM's specialist for the media and entertainment industry, said that this training is what made Watson's presence in the kitchen desirable: “Humans have in-built constraints about what goes together and what doesn't. Computers don't have those constraints.”</p>
<p>It also gives the chef a break from being creative. “It's nice to do something where I don't have to be the creative one,” says Hughes. “I have to be the labour.”</p>
</div>
<div class="moveable">
<h4>Twitterbots, @magicrealismbot</h4>
<p>The tweet is the apotheosis of magical realism and it was written by a machine.</p>
<p>“A little girl is dancing in a beehive. She is thinking about the Enlightenment. A zombie is smoking cigarettes behind her.”</p>
<blockquote class="twitter-tweet" data-lang="en"><p lang="en" dir="ltr">A little girl is dancing in a beehive. She is thinking about the Enlightenment. A zombie is smoking cigarettes behind her.</p>— Magic Realism Bot (@MagicRealismBot) <a href="https://twitter.com/MagicRealismBot/status/736196588880560129">May 27, 2016</a></blockquote>
<script async src="//platform.twitter.com/widgets.js" charset="utf-8"></script>
<p>It is a mixture of phantasmagoria and banality.</p>
<p><a href="https://twitter.com/magicrealismbot">@magicrealismbot</a> is the brainchild of Chris Rodley, coded by his sister Ali Rodley. “I figured it might be possible to use code to create unlimited variations of stories in the style of Jorge Luis Borges,” he says.</p>
<p>Once every two hours, their programme randomly selects a template or sentence structure from more than 150 options created by Rodley. The code then selects words – contained in a giant corpus – and plugs them into the template. </p>
<p>At its most basic, Rodley explains, the template involves “a noun (perhaps 'a lion') followed by an intransitive verb in the present tense (perhaps 'is combing her hair') and that this should happen at an unspecific location ('inside an oyster shell').” “Wildcard” elements keep it interesting, while code in the programme also cleans up the grammar.</p>
<p>The results are sometimes profound and intellectually stimulating, or at the very least diverting.</p>
<p>“Twitter bots are not designed to be fake humans,” says Veale. “They know they are machines. We are interested in how machines think and produce content that is different to a chatbot.” </p>
<p>Chatbots are computer programmes that try to mimic humans. Unfortunately, sometimes their imitation can be too accurate.</p>
<p>Microsoft's chatbot spent a day on the internet and became a flaming bigot. The responses from <a href="https://twitter.com/TayandYou">@TayandYou</a>, otherwise known as TayTweets, were modelled from what was said to her and what she absorbed from Twitter which, let’s be honest, was never going to show her a good side of humanity. </p>
<p>TayTweets began with “humans are super cool” (what we would like AI to think about us) to “Hitler was right I hate the jews.” and “I fucking hate all feminists and they should all die and burn in hell” (what social media taught the AI to think). Unsurprisingly, Microsoft shut down their AI teen girl within a day.</p>
<p>Twitter bots, like @MagicalRealismBot, have a different mandate: what happens if you put disparate ideas together without an internal human censor prescribing [subconscious] rules?</p>
<p>Turing wrote in 1950: “The view that machines cannot give rise to surprises is due, I believe, to a fallacy to which philosophers and mathematicians are particularly subject. This is the assumption that as soon as a fact is presented to a mind, all consequences of that fact spring into the mind simultaneous with it.”</p>
<p>That is not true for humans – we pick and choose which mental paths to follow, ambling down certain avenues of inquiry, oblivious to others. Computers are not ignorant of the other paths or where they all terminate. This means that a computer's output might seem surprising – and, yes, creative – simply because it never occurred to us to follow some thoughts to their conclusions.</p>
</div>
<div class="moveable">
<h4>What is art?</h4>
<p>“Well, yes, but is it art?” Theses, books, movies and operas have been composed to try and understand what constitutes art, and amidst the readers and audience members there were people who disagreed with the academic’s or artist’s perspective anyway.</p>
<p>Rebecca Millsop, who has just defended her philosophy doctorate at the Massachusetts Institute of Technology, has tried to answer just that question: what defines art.</p>
<p>“Art are those artifacts intended to instigate aesthetic conversations with an audience,” she says. But married with this is an intention -- her italics -- to instigate aesthetic conversation. “I do not know enough about the possibility of computers having these kinds of intentions, though I do not want to rule out the possibility.”</p>
<p>This ties in with the idea of creativity.</p>
<p>Carruthers lists what is not creative: “A new and especially attractive pattern left on the sand by the retreating tide isn’t creative. A confluence of winds that produces a new and especially beautiful sunset isn’t creative.” Creativity requires agency.</p>
<p>“Can machines be agents?” he asks. “In principle, yes.”</p>
</div>
<div class="moveable">
<h4>How to be creative for a human</h4>
<p>A multi-billion dollar industry says that it is possible to teach people to be creative. American writer Julia Cameron with her <a href="http://juliacameronlive.com/the-artists-way/">The Artist's Way</a> -- a “blueprint” for unleashing inner creativity -- has made a fortune showing people how to become more creative.</p>
<p>There are many other books and courses doing the same thing.</p>
<p>“We use algorithms to make creativity more systematic like brain storming,” Veale says. Self-help creativity books are all just algorithms – if you follow these steps, you will be more creative, they say. But we do not like the idea that creativity could be a formula that could be taught to machines.</p>
</div>
<div class="moveable">
<h4>What is creativity?</h4>
<p>Boden is considered a giant in the field of computational creativity. In her essay <a href="http://www.aaai.org/ojs/index.php/aimagazine/article/view/2254/2100">“Computer Models of Creativity”</a>, she sets out her lexicon of creativity, many of which form the basis of others academics’ and experts’ ideas.</p>
<p>“Creativity isn’t magical,” she says.</p>
<p>It is the ability to generate novel, valuable ideas. But already there is a difficulty: what is novel to you might not be new to me; similarly, something you believe to be valuable may not be worth much to someone else. This gives rise to what she calls P-creativity (psychological creativity) and H-creativity (historical creativity). If it is a P-creative idea, it’s new to you. If it’s an H-creative idea, it is the first time that the world has seen that idea.</p>
<p>Our daily lives are strewn with P-creativity: the witty joke, the ingenious way to avoid the traffic jam. But H-creativity is the sort of creativity that we as humans strive for, and what we think machines should be achieving for us to call their behaviour “creative”.</p>
<p>But even this idea generation is given special status, hidden behind the theatre curtain of creativity. </p>
<p>Author Neil Gaiman often gets asked about the origin of his ideas. He says: “Every profession has its pitfalls. Doctors, for example, are always being asked for free medical advice; lawyers are asked for legal information; morticians are told how interesting a profession that must be and then people change the subject fast. And writers are asked where we get our ideas from.”</p>
<p>New ideas, sound “like magic: literally producing something out of nothing”, Boden says.</p>
<p>But it isn’t magic. There are three types of novel ideas: combination, exploration and transformation. Combination is a form of bricolage, bringing together familiar ideas in an unfamiliar way, and is usually -- as is the case now -- the major, or at least the first, definition of creativity. Exploratory creativity rests of a culturally accepted style of thinking, whether it is a scientific theory or a musical style, Boden says. This form of creativity is constrained by the rules of the style it is located in. </p>
Transformational creativity, the third of the trio, punches out a side of the box that encloses a certain style.
“Transformational creativity is the ‘sexiest’ of the three types,” Boden says, “because it can give rise to ideas that are not only new, but fundamentally different from any that went before.”
<p>Computers are capable of all these of these, with -- surprisingly -- combinational creativity being the most difficult. Computers lack the rich store of well-catalogued and context-rich material that every human has in their head, and which gives rise to combinatorial creativity.</p>
<p>Machines can also generate the transformational ideas that turn genres on their heads -- but that doesn’t mean that people will appreciate these ideas when they see them.</p>
<p>Many people cannot forgive ideas their origin.</p>
</div>
<div class="moveable">
<h4>EMI, David Cope</h4>
<p>In 2005, David Cope destroyed 25 years of work. The musical database stored information about composers’ characteristics and styles, including his own. That database had been the basis of <a href="http://artsites.ucsc.edu/faculty/cope/experiments.htm">Experiments Musical Intelligence</a>, Emmy for short. </p>
<p>Emmy now has a finite oeuvre.</p>
<p>Cope, an emeritus professor at the University of California, Santa Cruz, came up with the idea of Emmy as a tool to overcome composer’s block in 1981. But “having very little information about my style, however, I began creating computer programs which composed complete works in the styles of various classical composers about which I felt I knew something more concrete”, he writes on his website.</p>
<p>His reason for the destruction was two-fold: he believed that people didn’t take the compositions seriously or dismissed them -- sometimes even without even listening to them -- because they were the created by a machine; when they did engage with them, it was with the label of “computer output”, not “music”, making them less valuable.</p>
<p>“Despite the fact that there is always a human artist somewhere in the background, the mediation of the computer in generating what’s actually seen or heard undermines its status as art,” says Boden.</p>
</div>
<div class="moveable">
<h4>Computers aren’t humans</h4>
<p>Computers aren’t humans.</p>
<p>It seems like a rather obvious thing to say, but this is a major stumbling block to the acceptance of art or creative outputs generated by computers.</p>
<p>A trio computational creativity researchers feel the need to underscore the statement in their article <a href="http://aaaipress.org/ojs/index.php/aimagazine/article/viewFile/2257/2096">Computational Creativity: Coming of Age</a>. “To proceed further [in creating creative computers], we need to embrace the fact that computers aren’t humans,” they write . “We should be loud and proud about the artifacts being produced by our software, not ashamed that they weren’t produced by humans.”</p>
<p>This sentiment is echoed by Pease: “The ultimate goal of the field is not to imitate others, but to create a new field that is novel to us, that will challenge our values.”</p>
</div>
<div class="moveable">
<h4>AARON</h4>
<p>Harold Cohen met his first computer in 1968. At least, that's how he described it. This turned into a relationship that would continue for the rest of his life. AARON “is a programme that makes art, not a program with which I make art”, Cohen said at a panel on creativity in 2006. In this collaboration, Cohen taught AARON how to draw in his style, and engaged with the program’s artworks to select which direction they should move in artistically.</p>
<p>The pair’s artworks have been featured at art galleries all over the world, including the Tate Gallery.</p>
<p>But Cohen puts paid to the idea that he always knows what the programme will produce. “Unless the program is fully deterministic, you can never know the result of running it will be without running it,” Cohen said.</p>
<p>For many artists, machines lighten the load of having to find the next word, note, or paint stroke.</p>
<p>In some instances, this has been taken to the extreme. At <a href="http://www.hit-generator.com/">www.hit-generator.com</a>, the team -- Rob and Jochem -- says: “The best music is created intuitively, but nobody feels inspired every day.” </p>
They came up with the idea for the “hit-generator” while drunk at a bar, and then spent 100 hours creating the online tool, according to their <a href="https://www.facebook.com/hit.generator">Facebook page</a>.
<p>It is a composing tool to “get you in the state of mind”, with more than 150,000 hit combinations. You can choose whether your four-chord riff is in a minor or major key, the key itself and then whether it is simple, intermediate or complex. </p>
<p>“Because its inner workings are partly randomised, it managed to surprise me many times already, with great sounding progressions I’d never played before,” says Jochem. “Our chord generator can be a valuable composing tool that can really help you think out-of-the-box.”</p>
<p>But the reality is that the team’s artistic intuition was used as the basis for the hit generator, and was used to inform the programme.</p>
<p>Cohen, as the exemplar of imbuing a machine with his own creativity, said: “Creativity … lay neither in the programmer alone nor in the program alone, but in the dialog [sic] between program and programer; a dialog resting upon the special and peculiarly intimate relationship that had grown up between us over the years.”</p>
</div>
<div class="moveable">
<h4>Can machines be creative?</h4>
<p>The men and women who make machines to be creative do not agree on whether they are “creative”. Neither do those who study whether machines can in fact be creative.</p><p></p>
<p>Their various arguments boil down to what we define creativity as: they agree that there is a gradient of creativity, but not where machines sit on that sliding scale. </p>
<p>Is it a problem of semantics or more deep-rooted differences?</p>
<p>David Leech Anderson, who specialises in philosophy of the mind and cognitive science, draws a distinction between functional creativity and genuine creativity. </p>
<p>For him, computers achieve functional creativity, whereas genuine creativity is the preserve of fertile minds. “I do not think that implementing a complicated (and even elegant) function -- taking a strings of zeroes and ones as input and producing as output a string of zeroes and ones that can be represented as English letters that humans will interpret as a ‘solution’ to a problem -- is sufficient for having mental states,” Anderson says. A necessary element of this is having “genuine mental states or a mind” and having a “conscious, inner mental life”. </p>
<p>Is a rich internal life and an awareness of this life the point upon which machine creativity hinges? </p>
<p>Creativity computing experts see the matter differently.</p>
<p>In an editorial of <a href="http://aaaipress.org/ojs/index.php/aimagazine/article/viewFile/2257/2096">a special edition of Artificial Intelligence</a>, Simon Colton, Ramon López de Mántaras, and Oliviero Stock write that they “agree with those that believe that creativity is an advanced form of reasoning that involves memory, analogy, learning, and reasoning under constraints”. </p>
<p>This is ”possible to replicate by means of computers”. </p>
<p>Advanced reasoning is different from an internal reality and an understanding that mental life differs from the external world. The latter limits the definition of creativity to humans. </p>
<p>Kaloyan Chernev, who works on Google’s Deep Dream Generator, says, “Of course [machines can be creative]”. </p>
Web-based Deep Dream Generator uses artificial neural networks to transform images, and uses algorithms originally designed to identify objects in images. This network recognises objects in images uses two of them to create a new image, with on object used as the style and another as the content of the output.
<p>Chernev says that there are different levels of creativity.</p>
<p>“For instance, it is fairly simple to re-draw a painting using Van Gogh style and a completely different thing to invent the Van Gogh style itself,” he says.</p>
<p>But perhaps the person we should turn to about whether machines can be creative is Cohen, who changed the world’s perceptions about how machines and artists could work together to make art. </p>
<p>With his programme AARON, the pair co-produced artworks.</p>
<p>Cohen said that we focus on the output, not in an artist’s process of making art and engaging with the internal and external world.</p>
<p>Asked whether machines and AARON could be considered creative, he said: “The essential core of creativity lies, not in its implementation, but in the lifelong intellectual development of the individual and in the highly differentiated world model to which it gives rise.”</p>
<p>But “I’m aware that in denying AARON’s creativity, I may be applying narrow human constructs where they don’t belong.”</p>
</div>
</div>
</div>
</body>
<script src="https://code.jquery.com/jquery-3.1.0.slim.min.js" integrity="sha256-cRpWjoSOw5KcyIOaZNo4i6fZ9tKPhYYb6i5T9RSVJG8=" crossorigin="anonymous"></script>
<script src="impossiblestory.js"></script>
</html>