RE: SIAI seeking seed AI programmer candidates

From: Michael Wilson (
Date: Wed Jun 02 2004 - 06:34:32 MDT

Giu1i0 Pri5c0 wrote:
> Very interesting Tyler. First, kudos for your "Becoming a Seed AI
> Programmer" article which is one of the very best job offered texts
> that I have ever seen.

The text is taken from the 'SoYouWantToBeASeedAIProgrammer' page on
the SL4 wiki, by Eliezer Yudkowsky.

> To business: reading the description of your project, and having
> some experience in managing large software development proejcts,
> I found myself trying to quantify your human resource needs and
> came up with a core staff of 50 software managers, architects,
> designers, coders, testers, plus cognitive scientists, mathematicians,
> hardware and communications experts, *insert here all current IT
> buzzwords*, and of course you also need support staff.

With respect,
a) you couldn't possibly know enough about the SIAI project to make
   a reasonable guess about the shape of the tasks involved and the
   amounts, types and grades of effort required to solve them.
b) it seems unlikely that you've worked on software projects that
   have successfully invented brand new technology, repeatedly,
   throughout the whole project while working within strict safety
c) on this project, everyone has to understand more or less everything
   that's going on. Keeping the bisection bandwidth down is critical.

> As part of the team you need some of the very best money can buy,
> people who would command a salary of 250k+/yr in the job market.

Nearly the whole team will need to be people like that. However even
if the cash is available to pay them those salaries, they shouldn't
be doing it for the money. This is the end of the world as we know it,
even if no-one else really feels that in their bones the programmers

> How good are your funding projections over the next few years?

It isn't possible to make reasonable projections because there are
too many high-risk high-payoff operations underway. We can only plan
for various possible outcomes.

> Yes of course you can use volunteers

Probably not, you need to think about this stuff nearly 24/7 to make
serious progress on it.

> Have you thought of intermediate marketable results that would look
> good on a business plan?

That's outside of the SIAI's remit as a charitable organisation.
Of course other parties may or may not be working on commercialisation
of related IP.

 * Michael Wilson


Yahoo! Messenger - Communicate instantly..."Ping"
your friends today! Download Messenger Now

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:47 MDT