Debates
- The field of Robotics has certain philosophical aspects to it. We
will
learn about this by means of class debates, which will occur at the
end of the term. A list of debate topics is included in your syllabus.
You will be asked to sign up for a topic and for the "pro"
or "con" position for that topic. You will have to prepare a 15-minute
argument for the idea you signed up for. You will deliver this argument
orally in groups in front of the class.
A. RSS Debate
A robotics debate team consists of two groups of two (sometimes 1),
debating on a resolution chosen in advance. Most teams will be 2-person
teams.
Debaters should prepare presentation materials (powerpoint slides
would
be appropriate). Before you do this, read the following suggestions
from Prof. Bruce Donald on giving a good
talk. Remember that your time is extremely limited - don't prepare
30 minutes of
great material, since we won't see most of it!
After the debates, debaters are required to turn in an
essay (1-2 pages) that argues their point in writing.
I. GENERAL DEBATE RULES
B. RSS Debate Rules - subject to (slight) change
Some basic readings will be assigned to the class covering the
debate topic. To win (or do well in the debate), each debater will
have to do outside reading and research on the topic in order
to have sufficient mastery to argue and rebut. This is especially
true for the rebuttal speaker, who should be able to respond
effectively to points made by the other team.
- Debates shall be organized as follows:
- Rules of Evidence
In debate, source citations of evidence must be stated the first
time a source is used.
- Rules of Evidence Authenticity
- Evidence must not be fabricated or distorted, in penalty of
losing
the debate.
Authenticity refers to:
- Fabrication: falsely representing a cited fact or
statement of
opinion as evidence; or intentional omission/addition of information
within quoted material.
- Distortion: misrepresentation of evidence or of citation
which
significantly alters meaning or content.
In your debate, you should be sure to describe the basic concepts
in adequate detail so that the jury (the class) can follow.
II. How to Win a Debate
In order to win, debaters will need to do a literature search and
some
extra reading (beyond the assigned papers) in order to assemble an
adequate armamentarium of technical material.
The best way to win an RSS debate is to make the most convincing
technical arguments in favor of your assigned position (pro or
con). For example, you could argue that while a particular theory of
AI claims to work on certain search problems, you have
proven that, for these problems, the search space is so constrained
that any search technique (including random) will perform
well! Or you could perform a careful complexity, soundness, or
completeness analysis to demonstrate your point. Or you could try
implementing the algorithm and show it performs well (or poorly).
Arguments by authority ("Rodney Brooks claims that X, and I believe
him"; "John Hopcroft, a Turing Award winner, believes neural networks
are bad so we should too") should not be used.
However, precise technical arguments ("Cybenko [give ref]
has proven that any continuous function can be modelled using a neural
network -- this means they have great expressive power") and technical
rebuttals ("Yes, but Cybenko's construction requires an exponential
number of hidden units -- great expressive power at great expense is
not useful!") are excellent debating points.
You must argue aggressively for your assigned position (pro or
con). Never say "Well, I've been assigned pro, but I really
agree more with con."
Finally, don't forget that the way you handle yourself during the
debate will influence your audience! Even the most coherent and
strong arguments can be undermined with poor presentation. This is
perhaps unfortunate (and less problematic with a "well-educated"
audience), but true. Be well-prepared, professional, respectful of
your opponent, and courteous to your questioners.
III. Grading of Debates
- All debaters (not teams!) will be given a grade based on their
performance.
- Debaters who do not show up for their assigned debate, come
unprepared to debate, or do not turn in an essay will receive an F.
- The jury (the class) will discuss the debate afterwards.
During this discussion, the jury may
cross-examine the debaters, and the debaters may cross-examine
each other. That is, during this time, the debaters can (and should)
continue to press their points.
- At the end of each debate the class will vote for the winner.
Your grade will not depend on this vote. Your grade will be based
on your argument, oral presentation, and written presentation as
evaluated by the course staff.
IV. Debate Topics
The resolutions chosen for the debates do not reflect a
judgement on the research. They are merely
chosen to be controversial and to stimulate discussion.
- Resolved, that deliberative robot
architectures are marvelously
clever search techniques whose running time and information
requirements make them completely inpractical for real-time
performance, unlike reactive robot architectures that will always give
the needed robot action in the required time.
Pro:
Con:
- Resolved, that building many sensors into
our robots as a way
of robustly solving problems is a habit that should be kicked.
Instead "just say know" (Matt Mason): use task-specific knowledge
to solve the problem.
Pro:
Con:
- Asimov's Three Laws of Robotics are
important and can and should
be designed into all robots that interact with people on a regular
basis.
Pro:
Con:
- In the near future we will be able to build
robots that are
as intelligent and powerful as humans, and will therefore have
the ability to take over from humans. We should therefore stop
conducting research into robotic intelligence.
Pro:
Con:
- In order to integrate robots effectively
into their environments it is better to develop monolithic,
all-knowledgeable and all-powerful robots, than swarms of many simpler
robots.
Pro:
Con:
- Little of value has ever come from robots
that do not imitate life.
The only valuable approach to creating robots is biologically-inspired,
following the principle that when it looks like a duck and it quacks
like a duck it must be a duck.
Pro:
Con:
RSS webmaster