From: Richard Loosemore (rpwl@lightlink.com)
Date: Thu Aug 24 2006 - 11:41:28 MDT
I have some specific, and quite serious answers to the issues you raise.
J. Andrew Rogers wrote:
>
> On Aug 23, 2006, at 3:30 PM, Richard Loosemore wrote:
>>
>> Alas I think that the $10M/year would not be enough for an Apollo
>> Project. 10M is only really about what a single decent startup would
>> need to get through seed stage.
>
>
> Only if a company does not know what they are doing. AI is not
> generally considered foundry work. $10M is a hell of a lot of money for
> a well-run software startup.
We could go into this in more detail, but I once costed the operations
of a combination software/hardware startup similar to the one that would
be involved here. There were about 40 engineering people, and some of
them were doing work that was robotics related. Remember: there would
be an absolute need to get peripherals involved (robot manipulators,
vision systems, auditory systems, etc), and that involves both buying
and playing with expensive hardware.
The burn rate for year one was projected between $16M and $22M.
Don't forget: I am not talking about some kind of shoestring internet
startup. I am addressing the issue of really building an AGI.
>> THE interesting question for me, however, is this: could a community
>> based project get its act together to do what NASA did for
>> spaceflight, but without government backing?
>
>
> No, because there are no usable metrics for incremental success.
> Effective communities do not materialize out of thin air. Even in the
> rarified atmosphere of this mailing list there is little agreement where
> it matters with regard to getting things done, and compared to the
> community at large we would appear to almost be in lockstep.
>
>
> Community-based AI research is the idea that having a clue is an
> emergent property of having enough people without a clue in one room.
> And I use "emerge" in the "magic happens" sense here. I would make the
> observation that most productive AI research seems to be done by people
> who have shown little interest in community-based AI research, which
> would suggest that the "community" aspect is quite irrelevant to their
> progress. It is not as though there are not plenty of community-based
> AI research projects in existence.
When I said "could a community based project [succeed]" I really meant
precisely, could it overcome the limitations of community based projects
at their worst (which is what you point to).
I have in mind a community that is able to act in a coordinated way,
with central direction, and in which the central directorate was more
than just some geek with an idea she got from AI 101, who gets a bunch
of dudes together to, like, hack an AI.
There are ways to get results that do not involve making gross
assumptions at the outset which cripple the project later on. While it
would not be possible to be completely agnostic about the approach, it
is quite feasible to be less ideologically committed and more receptive
to what the system data tell you to do.
As I say: could a community of people get together and do pieces of the
work under control of a central directorate that knew what the whole was
supposed to be, and which would have more than just a hack-and-fix
mentality.
These distinctions between ways of proceeding are subtle! They all look
the same to an outside observer. The magic is in the exact details.
>> We would not want governmet backing if we could help it, because the
>> conditions today are not the same as they were in the 1960s, and I
>> believe that the project would be pork-barrelized today where it was
>> only partially true back then.
>
>
> This statement seems a bit baseless and arbitrary. It might be true,
> but it is not obvious.
Let's not get into politics. It is stupidly obvious to some who do a
lot of reading and who have a lot of experience. Let's just leave it at
that. Not an important issue.
>> I'd be prepared to initiate such a project right now if the startup
>> funds were available, because I believe it could actually work. And
>> in case I am not being clear, I don't mean just another advocacy group
>> or think tank, I mean a project to get concrete things done.
>
>
> Starting a concrete project requires having a concrete and thorough
> development plan at the outset or it will be a dead letter. This sounds
> like forming a committee that will select the advisory committee to the
> project planning committee that will decide what the project actually
> will be doing. In other words, several steps too early to really expect
> anyone else to be interested or involved.
>
> If you need a community to tell you exactly what you should be doing,
> you might as well give up. And if you know exactly what it is you
> should be doing, what is the critical value of the community exactly?
Okay, I am guilty of making what seemed like an offhand comment.
You could not be more wrong about the "forming a committee that will
select the advisory committee", etc etc. ;-).
I made the comment because I have enough specific goals and detailed
designs to put a gang of programmers to work within days. I don't need
a community to tell me exactly what I should be doing, I need a
community to DO exactly what I tell them to do, and no messing.
Okay, that sounds a little dictatorial. I speak half in jest, because
your comment made me laugh. I don't think I have given the impression
on this list that I like waiting for committees to tell me what to think
or do.
The critical value of the community would be to (a) do some hard work,
and (b) do some serious research about fine details. Heck, just the
usual mix you would get in a company of 40-odd people.
The problems would be: (a) you would need some kind of central
organization because the hardware would be serious, and could not be
completely distributed, and (b) the personal attitudes in the community
would be critical, and perhaps not appropriate for the work to be done:
they couldn't be a collection of mavericks who hated the idea of being
given chunks of work to do.
Anyhow, we are debating whether this is possible almost as if we were
talking in a vacuum: any minute now Ben is going to check in from his
vacation hideawy and point out that what I am describing IS what
Novamente is already doing. Novamente is partly community based, but
has a strong central design plan. What I have in mind is (with all due
respect to Ben) just something on a larger scale, and with a more (how
can I put it?) neutral and empirical aspect. My focus is on building a
specific set of tools first, then using those to take measurements that
would solve some (carefully delineated) design features.
The problem is that the specific design I have in mind would follow a
path that is hard for me to describe without dumping an 80,000 word
thesis into this message. That design path is, in some senses, quite
ecumenical (i.e. it accommodates quite a variety of different ideas
about AGI), but in other senses it could be taken as (or misunderstood
as) quite radical and non-mainstream. It stands in the weird and
difficult position of being an approach to AGI that, before it was
built, would irritate the heck out of the mainstream folks (because it
challenges some deep assumptions that they are emotionally committed to,
and which they really find difficult to comprehend or talk about), but
which, *after* it was built, would probably be said to be quite
conservative and cautious and obvious.
I am not going to start talking about that design, because I am no
longer interested in wasting my time fighting battles across paradigm
boundaries (any more than I want to waste my time herding cats or trying
to get talkative 7-year-olds to bed on time).
On balance, I think a community approach probably would not work. I
think it would need a company, and real funding.
Richard Loosemore
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:57 MDT