]> git.zerfleddert.de Git - micropolis/blob - src/notes/Beyond-Intelligent-Machines
automatically detect byteorder of X server
[micropolis] / src / notes / Beyond-Intelligent-Machines
1 Date: Tue, 24 Nov 92 22:25:59 -0500
2 From: ben@cs.UMD.EDU (Ben Shneiderman)
3 To: bam@cs.cmu.edu, ben@cs.umd.edu, callahan@cerc.wvu.wvnet.edu,
4 hopkins@bongo.garnet.cs.cmu.edu, weiser.pa@xerox.com
5 Subject: Re: more pie menus!
6
7 I couldn't resist sending you all this latest essay which is
8 destined for IEEE Software...some readers expect it to generate
9 some strong responses...Ben
10
11
12 Beyond Intelligent Machines:
13 Designing Predictable and Controllable User Interfaces
14
15
16 Ben Shneiderman November 24, 1992
17
18 University of Maryland, College Park, MD 20742
19
20 Professor, Department of Computer Science,
21 Head, Human-Computer Interaction Laboratory at the
22 Center for Automation Research &
23 Member, Institute for Systems Research
24
25
26 Who's in control?
27
28 An important shift is occurring from the old vision of computers
29 as 'intelligent' to a new vision based on predictable and controllable
30 user interfaces that depend on direct manipulation of objects and actions.
31 Appropriate metaphors and terminology are important since they shape
32 the thoughts of researchers, designers, managers, congress-people,
33 journalists, etc. Most of us have learned the importance of gender
34 neutral terminology and similarly I have been strongly opposed to
35 suggesting that computers are 'intelligent' or 'smart' for several
36 reasons:
37
38 1) Limits to Imagination
39
40 I think we should have much greater ambition than to make a computer
41 behave like an intelligent butler or other human agent. Computer
42 supported cooperative work (CSCW), hypertext/hypermedia, multi-media,
43 information visualization, and virtual realities are powerful
44 technologies that enable human users to accomplish tasks that no human
45 has ever done. If we describe computers in human terms then we
46 run the risk of limiting our ambition and creativity in the design
47 of future computer capabilities.
48
49
50 2) Predictability and Control are Desirable
51
52 If machines are 'intelligent' or 'adaptive' then they may become less
53 predictable and controllable. Our usability studies show that users
54 want feelings of mastery,
55 competence, and understanding that come from a predictable and
56 controllable interface. Most users seek a sense of
57 accomplishment at the end of the day, not the sense that this
58 'intelligent' machine magically did their job for them.
59
60
61 3) Human Responsibility
62
63 I am concerned that if designers are successful in convincing the users
64 that computers are intelligent, then the users will have a reduced sense
65 of responsibility for failures. The tendency to blame the machine is
66 already widespread and I think we will be on dangerous grounds if we
67 encourage this trend.
68
69
70 4) Machines are not People AND People are not Machines
71
72 I have a basic philosophical objection to the suggestion that machines
73 are, or can ever be, intelligent. I know that many of my colleagues are
74 quite happy to call machines intelligent and knowledgeable, but I prefer
75 to treat and think about machines in very different ways from the way I
76 treat and think about people.
77
78
79 The lessons of history
80
81 While some productive work has been done under the banner of
82 `intelligent', often those who use this term reveal how little they
83 know about what users want or need. The users's goal is not to
84 interact with an 'intelligent' machine, but to create, communicate,
85 explore, plan, draw, compose, design, or learn. Ample evidence
86 exists of the misguided directions brought by 'intelligent' machines:
87
88 - natural language interaction seems clumsy and slow compared to
89 direct manipulation and information visualization methods that use
90 rapid, high-resolution, color displays with pointing devices. Lotus HAL
91 is gone, AI INTELLECT hangs on but is not catching on. There are some
92 interesting directions for tools that support human work through
93 natural language processing: aiding human translators, parsing
94 texts, and generating reports from structured databases.
95
96 - speech I/O in talking cars and vending machines is gone.
97 Voice recognition is fine for handicapped users plus special situations,
98 but doesn't seem to be viable in general office, home, or school
99 settings. Our recent studies suggest that speech I/O has a greater
100 interference with short term and working memory than hand-eye
101 coordination for mouse menu selection. Voice store and forward,
102 phone-based information retrieval, and voice annotation have great
103 potential but these are not the 'intelligent' applications.
104
105 - adaptive interfaces are unstable and unpredictable, often leading
106 users to worry about what will change next. I see only modest chances
107 for success in user modeling to recognize the level of expertise and
108 revise the interface accordingly - can anyone point to successful
109 studies or commercial products? By contrast, user controlled
110 adaptation through control panels, cruise control for cars, and
111 remote controls for TV are success stories. While algorithms to
112 deal with dynamic
113 issues in network or disk space management are needed, the task domain
114 and user interface issues of the application program
115 should generally be under direct user control.
116
117 - Intelligent CAI (Computer Assisted Instruction) only prolonged the
118 time (compared to traditional CAI) until the users felt they were the
119 victims of the machine. Newer variations such as Intelligent Tutoring
120 Systems are giving way to Interactive Learning Environments where
121 students are in control and actively creating or exploring.
122
123 - intelligent talking robots with five-fingered hands and human facial
124 features (quaint fantasy that did well in Hollywood but not in Detroit
125 or elsewhere) are mostly gone in favor of flexible manufacturing systems
126 that enable supervisors to specify behavior with predictable results.
127
128
129 It seems that some designers continue to ignore this historical pattern
130 and still dream of creating 'intelligent' or 'smart' machines. It is an
131 ancient and primitive fantasy, and its seems most new technologies must
132 pass through this child-like animistic phase. Lewis Mumford identified
133 this pattern (Technics and Civilization, 1934) when he wrote about the
134 Obstacle of Animism: 'the most ineffective kind of machine is the
135 realistic mechanical imitation of a man or another animal...for
136 thousands of years animism has stood in the way of...development.'
137
138
139 An alternate vision
140
141 My point in this essay is not merely to counter a popular design
142 philosophy, but to offer a new vision that is more in harmony with what
143 users want. I believe that the future will be filled with powerful, but
144 predictable and controllable computers that genuinely serve human needs
145 (Designing the User Interface: Strategies for Effective Human-Computer
146 Interaction, Second Edition, Addison-Wesley Publ. Co., Reading, MA, 1992).
147
148 In this vision of predictable and controllable (PC) computing,
149 the promising strategies are rapid,
150 visual, animated, colorful, high resolution interfaces built on
151 meaningful control panels, appropriate preference boxes,
152 user-selectable toolbars, rapid menu selection, easy to create macros,
153 and comprehensible shortcuts. These enable me to specify rapidly,
154 accurately, and confidently how I want my email filtered, what documents
155 I want retrieved and in what order, and how my documents will be
156 formatted.
157
158
159 Our Human-Computer Interaction Laboratory has applied these principles
160 to information visualization methods that give users X-ray vision to see
161 through their mountains of data. Treemaps enable users to see (and
162 hear) 2-3000 nodes of hierarchically structured information by utilizing
163 every pixel on the display. Each node is represented by a rectangle
164 whose location preserves the logical tree structure and whose area is
165 proportional to one of its attributes. Color represents a second
166 attribute and sound a third (B. Johnson & D. Turo, Improving the
167 Visualization of Hierarchies with Treemaps: Design Issues and
168 Experimentation, Proc. IEEE Visualization '92). Treemaps have been
169 applied to Macintosh directory browsing (Figure 1), in which area could
170 be set to file size, color to application type, and sound to file age
171 (our TreeViz application is available from the University of Maryland's
172 Office of Technology Liaison, (301) 405-4210). When users first try
173 TreeViz they usually discover duplicate or misplaced files, redundant
174 and chaotic directories, and many useless files or applications. Other
175 applications include: stock market portfolio management, sales data,
176 voting patterns, sports (48 statistics on 459 NBA players, in 27 teams,
177 in four leagues), etc.
178
179
180 Dynamic queries allow rapid adjustment of query parameters and immediate
181 display of updated result sets. These animations enable users to
182 develop intuitions, discover patterns, spot trends, find exceptions, and
183 see anomalies. The Dynamic HomeFinder prototype (Figure 2) allows users
184 to adjust the cost, number of bedrooms, location, etc. and see points of
185 light come and go on a map to indicate a matching home. Users execute
186 up to 100 queries/second (rather than one query per 100 seconds)
187 producing a revealing animated view of where high or low priced homes
188 are found, and there are no syntax errors. Clicking on a point of
189 light brings up a description or image (videotape available, or for
190 an empirical comparison with a natural language system, see
191 Williamson, C. and Shneiderman, B., The Dynamic HomeFinder: Evaluating
192 dynamic queries in a real-estate information exploration system, 1992
193 ACM SIGIR Proceedings).
194
195 Dynamic queries are very effective when a visual environment such as a
196 map, calendar, or schematic diagram are available, but they can be
197 easily applied with standard text file output (Figure 3). Dynamic
198 queries exemplify the future of interaction; You don't need to
199 describe your goals, negotiate with an intelligent agent, and wait for
200 a response, you Just Do It! Furthermore, dynamically seeing the
201 results enables you to explore and rapidly reformulate your goals in
202 an engaging videogame-like manner.
203
204
205 Open problems in information visualization include screen organization,
206 widget design, algorithms for rapid search and display, use of color
207 and sound, and strategies to accommodate human perceptual skills.
208 We also see promise in expanding macro makers into the graphical
209 environment with visual triggers based on controlled replay of
210 desired actions - the
211 general idea is Programming in the User Interface (PITUI) to
212 Do-What-I-Did (DWID).
213
214
215 I want to encourage the exploration of new metaphors and visions of how
216 computers can empower people by presenting information, allowing rapid
217 selection, supporting personally specified automation, and providing
218 relevant feedback. Metaphors related to controlling tools or machines
219 such as driving, steering, flying, directing, conducting, piloting,
220 or operating seem more generative of effective and acceptable
221 interfaces, than 'intelligent' machines.
222
223
224 A scientific approach to user interface research
225
226 Whether you agree with the design philosophy in this essay, and
227 especially if you disagree, I hope that you will add to our scientific
228 knowledge by conducting well-designed empirical studies of learning
229 time, measuring performance time for appropriate tasks, recording error
230 rates, evaluating human retention of interface features, and assessing
231 subjective satisfaction. There's much work to be done to make
232 computing accessible, effective, and enjoyable.
233
234
235 Acknowledgements: This essay was prompted by the discussion between
236 Mark Weiser and Bill Hefley, stimulated by lively email and personal
237 discussions with Paul Resnick, Tom Malone, and Christopher Fry at MIT,
238 and refined by comments from Catherine Plaisant, Rick Chimera, Brian
239 Johnson, David Turo, Richard Huddleston, and Richard Potter at the
240 Human-Computer Interaction Lab at Univ. of Maryland. I appreciate Bill
241 Curtis's support for this vision. Thanks to all.
Impressum, Datenschutz