home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!elroy.jpl.nasa.gov!swrinde!mips!darwin.sura.net!haven.umd.edu!mimsy!hendler
- From: hendler@cs.umd.edu (Jim Hendler)
- Newsgroups: comp.ai
- Subject: AI In the Year 2025 - Responses
- Summary: Pretty Long message
- Message-ID: <59282@mimsy.umd.edu>
- Date: 28 Jul 92 18:56:22 GMT
- Sender: news@mimsy.umd.edu
- Organization: U of Maryland, Dept. of Computer Science, Coll. Pk., MD 20742
- Lines: 374
-
-
- As promised, I have collected responses to my request for predictions
- on AI (and CS) in the year 2025. I deliver a lecture on the subject
- tonight, and I profusely thank all of you who contributed. Below are
- the ones that I will be using in the lecture. They are about 80% of
- those I received, many others that had good ideas, but weren't in the
- form of short predictions had to be dropped.
-
- I look forward to sealing this list in a time capsule and hope I might
- be around in 2025 to see which of it has come true...
- -Jim Hendler
-
- <apologies if I got name or affiliation wrong, I processed a lot of
- these in a short amount of time>
-
- John Josephson, Ohio State University
-
- Domestic robots able to do routine cleaning, including laundary and
- kitchen cleanup. I predict that by 2025 these robots will be the
- basis for a multi-billion dollar industry.
-
- Self-guided automobiles. I predict that by then we WILL still have
- private automobiles, BUT that driving skills as we presently know them
- will not be required. It is not clear how much "general AI" the autos
- will have - e.g., the ability to read road signs - as opposed to
- special-purpose smarts tuned to roadways made purposely easy for them
-
- I predict with high confidence that people will no longer have to
- memorize phone numbers by 2025. .. jj
-
- Bill Armstrong, Univ. Alberta, Canada
-
- Robots based on adaptive logic network (ALN) technology will be trained to
- perform many complex tasks which today require humans, e.g. garbage recycling.
-
- Kenrick Mock, UC Davis
- I foresee a larger scale and more unified view of AI.
- Genetic algorithms, neural nets, case-based reasoning, etc.
- all share a common mathmatical foundation. While currently
- limited to toy domains, in the future these techniques will
- apply to complex problems involving thousands or millions
- of features.
-
- Mark Shanks, Honeywell Inc.
-
- I forsee being able to purchase a "friend" the way you buy typical
- software. This AI-based (Eliza-like) program would be customizable
- as to sex, interests, age, etc., and would include a high-resolution
- animated image. Communication would be vocal; no keyboard. And its
- a scary thought..........
-
- Mark Kantrowitz, Carnegie-Mellon U.
-
-
- 1. Transportation: AI will form part of the implementation of new features in
- automobiles, such as automatic parallel parking, tailgating alarms,
- and route planning.
-
- 2. Entertainment: AI will run the minds of characters in interactive
- entertainment applications based on interactive fiction and virtual reality
- technology.
-
- 3. User Interfaces: AI will be involved in user interfaces that recognize and
- generate text, speech, handwriting, and gestures. In addition, AI
- technologies will also be able to recognize and generate the emotional and
- stylistic content of such communication.
-
- 4. Information Management: KnowBots will manage the user's personal information
- database, must like a butler or secretary. Such information includes the
- user's calendar, phonebook, bank accounts and bills, and inventory.
-
- 5. Email for the masses: Electronic multimedia mail will be commonplace, as
- telephones, computers, television sets, laser printers and fax machines merge.
- Every household will have an inexpensive telephone instrument capable of
- transmitting speech, text, still images and live video. KnowBots, such as
- described in 4 above, will serve as the interfaces to such systems.
-
-
- Robi Bisesarji
-
-
- by 2025 computers will start writing poetry, not unlike that of keats,
- shelly and rabindranath tagore.
-
- Tim Finin, (Chair, Computer Science), Univ. of Maryland, Baltimore County
-
- In the year 2025 computer keyboards will have ended up where ever all
- the Underwood's are now. Spoken language and pen-based graphics will
- have completely replaced the keyboard as an i/o device on all computers.
-
- In the year 2025, the computer, the telephone network and television
- will have completely merged.
-
- In the year 2025, mass-market and even *-market newspapers, magazines,
- and books will have disappeared. Individualized and custom tailored
- "documents" will be downloaded to your local system (at home or the
- office) for viewing, archiving or printing.
-
- David Demers, UC San Diego
-
- Voice recognition systems will be very good, widely implemented, and
- will no longer be considered "AI".
-
- The world's best chess, checkers and backgammon players
- will be programs. The world's best bridge players will
- be human.
-
- Dave
-
- Afzal Ballim, University of Geneva, Switzerland
-
-
- ..Will be the most overused advertising slogan - "Now with new improved AI,
- washes even cleaner than before."
-
- Wim Jansen, Tech. Univ Eindhoven, Netherlands
-
- By the year 2025 decisionmakers of large corporations will trust their
- expert system over their human advisors.
-
-
- John Morrill, Ohio State Univ.
-
- I predict that in 2025, the home workstation will handle most of the household
- finances. It will recieve bills, project needed cash, send payment orders and
- transfer funds between accounts, all with out the need of human supervision.
-
- Drago Indjic, Imperial College, London
-
- Attentional (filtering) systems: the "coctail party speech processors"
- capable of focusing attention to a single object - hence avoid
- (hierarchical) search and the curse of dimensionality.
-
- Being an extension of adaptive filtering (and strikingly close to
- radar systems), it could be easily developed as a subsystem for vision
- or hi-dimensional control.
-
- Of course, there will be many technologicaly-driven developments, and
- (data-level at least) parallel computers and wireless communications
- will be well integrated into sociotechnological environment. A nice
- example is a movie "Until the end of the world" by Wim Wenders, quite
- plausible sci-fi forecast, IMHO.
-
- Stephen Aylward, McDonnell Douglas Corporation
-
- Neural Network Researcher :
- In the year 2025, we will be able to model the human brain.
-
- Neurobiologist :
- In the year 2025, we will be able to model a single neuron.
-
-
- Patrick Jost, TRW
-
- Natural Language Processing that works-passes the Turing Test, does
- not depend solely on keywords or brittle parsers, carries on
- conversations, does translations, searches based on content; the
- limiting factor will be knowledge representation.
-
- Clark Quinn, U. New South Wales, Australia
-
- In the year 2025, there will exist enjoyable explorable programs that
- will coach people through just about any topic they care to learn
- about, including creating enjoyable explorable programs....
-
- Jim King, NCR
-
- Tapping the resources of the mind through artificial stimulation and
- embedded neuro-devices will be state of practice in 2025. Commercial
- offerings will include chemical, mechanical (micro machines), neural
- and biological approaches to enhancing decision making within the
- human brain - and not through external replication of the brain's
- processes as today.
-
- Anonymous
-
- Formalizing common-sense views of human behavior will allow a new maturity
- across all the social sciences.
-
-
- James Franklin, U. New South Wales, Australia
-
- Image processing will be so accurate that cancer cells will be detected
- one by one, and zapped. Cancer patients will stay alive indefinitely.
-
-
- Basile Starynkevitch, C.E.A., France
-
- in 2025 the chess world champion will be a computer AI program.
- In 2025 the world Go champion will still be a human.
- In 2025 automatic technical report translation will be common and usable.
- In 2025 garbage collection in wealthed cities will be done by AI based robots.
-
- T. Mark Ellison, University of Edinburgh
-
-
- Adaptive learning architectures will break down the current distinction
- between neural nets and symbolic learning methods.
-
- Jordan Pollack, Ohio State University
-
- The best AI will continue to be where the cashflow is greatest, in
- video-game entertainment where artificial lifeforms populate
- artificial reality.
-
- Everyone will have a compute server equivalent to a CM5
- in their basement, next to the water heater and furnace,
- but it will be inadequate for anything besides balancing the
- checkbooks.
-
- Tiny robots, about the size of cockroaches(expensive) or mice (cheap)
- will be everywhere doing special purpose domestic chores, like
- vacuuming, dusting, dishwashing, and folding laundry.
-
- "Microsofts" will be the word for 3 by 5 foot color lcds which hang
- on the wall and display changing artwork based on neural network
- analysis of the mood of voices in the room.
-
- Televisions for the rich will have AI watchers built in, with enough
- disk memory to delay a program 15 minutes and cut out the commercials.
-
- AI will be used to control the allocation and scheduling of piece work
- to distributed home workers, because pollution is too great to go
- outside.
-
- Doug Whitehead, Corp. for Open Systems
-
- I believe HCI will grow in importance, through extentions of
- the Xerox PARC Ubiquitious Computing project. We will have an
- INVISIBLE monitor located on our body that communicates with
- our surroundings, announcing our intent, and retriving
- information for us. It will learn from the owner and become
- an extention of that owner.
-
- An example of an unobtrusive interface with the owner:
- project icons, or visualizations into the eye, so they seem
- to float in space in front of the person (similiar to the
- "private eye" display that exists today), and watch for
- eye movements to detect preference.
-
- It will make us more productive, but will also
- force us to acknowledge that a good percentage of
- our time is spent; piddled away in thought, or puddering
- around the office.
-
- Eric Iverson, New Mexico State Univ.
-
- By 2025 the machines we have created will define what is intelligent
- and we won't qualify.
-
- By 2025 machines will be so intelligent that they will be able to
- delude us into thinking that we solved a problem by ourselves.
-
- By 2025 machines will be so small that they will be able to fit into a
- briefcase. Assuming, of course that the briefcase in question is only
- visible under heavy magnification.
-
- Glenn Holiday, Computer Sciences Corp.
-
- In 2025, we will have no big, smart programs. We will have lots
- of small, dumb programs with just enough AI to survive. They will be
- specialist agents, each taking over a specific load now done by humans.
-
- In 2025, people will still be debating whether we will have
- built something truly intelligent by 2050.
-
- R. Peter Bonasso, Mitre Corp.
-
- Just like your present day Mac. You order it with whatever
- sensor/actuator options you want and when it arrives you take it out
- of the box, slip in the applicatons disk, boot it up and "Hi, I'm
- Genie. Your wish is my command." -- away you go.
-
- Why? Because of advances in robot architectures -- a truly
- multitasking robot OS which has drivers for various sensors and
- actuators and layers capabilities from safe negotiating of unknown
- environments to planning a meal and carrying out the plan. Because of
- active perception techniques which allow robots
-
-
- Thurstan Felstead, Brunel University, U. K.
-
-
- "AIs (Agent Intelligences) directed from within a Total Immersion
- Information Workspace accessing a WAN (World Area Network). Time Line:
- 1980s - The Data Age, 1990s - Transition, 2000s - The *TRUE*
- Information Age."
-
-
- Haakon Styri, Norwegian Telecom Research
-
- My phone answering machine will try to behave in an intelligent
- manner, but I don't know if I'll appreciate it. ("The person you want
- to reach is in the shower, please press 1 if your message is really
- urgent...")
-
- Harry Erwin, TRW
-
- The vulnerability of capitalist economies to the dynamics that
- underlie panics and market swings will be eliminated by 2025 as
- effective tools for predicting and controlling discrete chaos are
- identified.
-
-
- Bruce Krulwich, Nortwestern Univ.
-
-
- In the year 2025 appliances will be able to figure out our habits and
- preferences, and make it easier for us to use them in the ways we like
- to. Our lights will know when we want them on, our stereos will know
- what stations we like to listen to, our refrigerators will know what
- we like to drink at different times, and so on.
-
- Ben Grosof, IBM
-
- 1) Consumers will control even appliances as humble as ovens and
- washing machiness by voice dialogue.
-
- 2) Newly built houses will actively and autonomously co-ordinate
- lighting, climate control, security/alarm functions and be
- re-programmable by voice dialogue in person or remotely.
-
- Mark Slack, Mitre Corp.
-
- Robotics reaches a state where routine jobs are more cost effectivly
- preformed by robots (burger king, steel workers, high way
- maintainence, etc). The resulting unemployment rate exceeds 25% as
- the displaced workers lack employable education levels. Massive riots
- ensue as the government has not anticipated the social impact of
- robotic automation. Society is further polarized between the haves
-
- Northrop Fowler III, Rome Laboratory
-
-
- Today some of us view computers as indispensible PARTS of our everyday
- environment (work, play, entertainment, etc). The logical
- extrapolation is that in XX years (25?) we will have crawled into the
- computer, so's to speak. That is, the computer will become virtually
- (!) indistinguishable from our environment (or vice versa!). It will
- manage almost every facet of our physical and mental well-being,
- providing instant access to the "climate" of our choice. And the
- principal key to this evolution in control over our environment is:
- AI!
-
- Sorry, gotta go; have a 1960's appointment down on Holodeck 3. Today
- I get to drive that new 1957 Chevy convertible (yes, with the blonde!)
- I dreamed about as a kid!
-
-
- John Frank, Rome Laboratory
-
- I predict that AI will be to the point that it will be used to
- generate the designs and code for the next generation of AI based
- systems across, as a minimum, all systems involving well defined limits
- and operations.
-
- David Dobbs, DD Wyndham, Canada
-
- Intelligent devices using the technology of multiple paradigms combining
- neural networks, genetic algorithms, fuzzy sets, expert systems and dynamic
- systems ("chaos") eclipse the old-fashioned 'mono-culture' AI approach in
- capabilities. Robots using a combination of DS, GA and FS are capable of
- self-planning far more complex activity than previous robots.
-
-
- Jim Hendler, University of Maryland
- There is no illusion more dangerous than the belief that the progress
- of science is predictable.
- Freeman Dyson, 1992
- --
- Jim Hendler
- Department of Computer Science
- University of Maryland
- College Park, Md 20742
-