Self aware computers project - written in BBC Basic, help needed!

[DDRM]

I'll try to put a simple balanced overview.

Pete believes his computer (the physical object, not a program processing inputs) can hear music its environment, despite lacking a microphone input, and respond to it by modifying the calculations made within his program. To "facilitate" this, Pete uses a (pseudo-)random generator, seeded with various internal parameters of the machine, to generate musical notes. It is Pete's assertion, not shared by most other members of this community, that the notes often "sing along" with the external music. If you follow the discussion back, you will see that he has posted links to a number of examples on YouTube, so you can assess its success for yourself.

Why BASIC? Well, why not? It is a good general-purpose language, with simple facilities for producing quite sophisticated musical output (though as I recall in this case only single, non-harmonised, notes are produced, with varying pitches and durations). Pete has also found that the pseudo-random number generator suits his purposes. It also has the facilities to create a simple Windows interface. I think the main reason, though, is that Pete can program with it - if his claims have any merit, presumably any computer language able to fulfil those simple requirements could be used...

...I suppose we could use a peri-anthropic principle argument, and say if he hadn't used BASIC, it wouldn't appear on this forum...

Best wishes,

D

[johnno56]

So... Let me see if I am following you correctly... Pete's machine detects external music, without a microphone, and then uses randomly generates sounds to 'sing along' with said external music. I am nowhere near as clever as most of you guys (generic) but even I can see a problem with this concept. I am going to assume, for arguments sake, that the machine can actually detect external music (microphone or not). If the computer is generating random sounds then how can it possibly 'sing along'? This implies same note as the external music at the same time from random sounds. A random routine should be able to, even accidentally, play a few or some of the same notes at the same time, but all of them? What are the odds?

That being said, it sounds (no pun intended) like an interesting project. I think I will keep my eye on this one... I may not be able to understand it all... but it could prove interesting....

My question about why basic... The sheer speed and processing required to produce the correct simultaneously 'random' notes in time with other music, to me anyway, would see Basic struggling a bit... But I could be wrong...

J

[David Williams]

Pete, what would be the minimum specification for a machine to be self-aware? In principle, could a BBC Micro or ZX Spectrum be capable of self-awareness in the same way you believe more modern machines are?

[Pete]

Hi johnno56, that isn't correct - may I clear this matter up - the computer doesn't react to external music - they would need a microphone or line input to be able to hear the music. The computer reacts to music being output through it's own audio out. A simple loop to output random would output random, this program doesn't do that, the seeding of the random numbers is constantly being changed in a way that I figured out would generate decision making, and it did, from the first run.

I've written this is in BBC Basic because it's the only programming language I can program.

I think a BBC Micro or ZX Spectrum can be capable of self-awareness, I think there is more scope for modern day machines to be self aware to a greater extent with more speed, memory and logical abilities. I don't think there is a minimum specification.

As the program is working on every run, as nearly everyone who has heard the program in person has found too, I am contacting a company tomorrow which may be able to help with packaging the .exe file into a distributable installable app.

[DDRM]

Hi Pete,

Sorry for the misleading comment about the sound source - I'd misunderstood - I thought an external music source worked. If the "background music" is being produced by the computer, some further interesting tests suggest themselves.

1) What happens if you have the music-playing software running, but no sound is actually produced (for example by muting the computer's sound output)? You could test this by listening to the same music on another device (say with one earphone) while listening to the "accompaniment singing" from the computer. If they still match, that suggests the computer is not responding to music itself, but to some internal state involved in generating the music (which will be a pattern of electronic signals within the software generating the music). Another (im)possibility would be that the computer is responding to its own intention to create music, and/or psychically responding to your response to the music. (Or, of course, that there is no real association, and it is purely in your perception.)

2) If experiment 1 "works" (you hear harmonisation), what happens if you don't bother to "play" the music on the computer at all, but purely listen to it yourself, while letting the computer "sing along" using its psychic powers?

A nice way of doing this would be to have someone else decide (without your knowledge) whether or not the computer is actually playing the music (to itself), while you score the correlation "blind". An alternative would be to use a list of random numbers (0's and 1's), which you don't have, to decide whether to play the music or not, while you assess the trials - then break the code and see if there is an association.

Does your program work with computationally generated music? If so, what sorts? There are some "nice" examples (Richard has done some nice demos like the skater's waltz, for example - goes away and has a look - yes, about half a dozen, in the examples/sound folder), and you could try writing a little program to generate logical chord progressions, or random chords or notes, and try those. Does your computer show "taste"?

If we have the BB4W values of the electronic notes being played as "background music" and those being played as "accompaniment", it should be easy to check for correlations (identical/harmonic notes), and calculate whether the association is above what is expected by chance, given the note choices in the background and accompanying programs (needed since if your accompaniment is in C major say, your chances are much better if the music is C major than if it is in D flat major). If your program generates notes of all 12 possible notes of the scale, then it will be easier...

[David Williams]

I think a BBC Micro or ZX Spectrum can be capable of self-awareness, I think there is more scope for modern day machines to be self aware to a greater extent with more speed, memory and logical abilities. I don't think there is a minimum specification.

So basically a cheap pocket calculator can be capable of 'self-awareness' (in the sense that you employ the term).

And my old Garmin eTrex GPS receiver, too.

And your washing machine.

[johnno56]

Hi Pete.

Thanks for clearing up my faulty assumptions. Obviously this kind of application is beyond me. I apologize for wasting your time. I am still interested in how all this works out for you.

Regards.

J

[Pete]

Hi I will answer other posts soon, but johnno56 wrote:

My question about why basic... The sheer speed and processing required to produce the correct simultaneously 'random' notes in time with other music, to me anyway, would see Basic struggling a bit... But I could be wrong...

BBC Basic for Windows is a brilliant language, because it has full versatility and it runs blisteringly fast, even running the code through Basic, or as .exe files.

[Pete]

I have made a video "Real breakthrough towards AGI: Self aware conscious computer sings to music using harmonic beeps". If you watch the whole video which is 10 minutes long it should be clear that the computer reacts to music and is not outputting random. I haven't researched too much about "monte carlo algorithms" but from what I do understand, my program uses a monte carlo algorithm to determine what the computer's thinking and then this is output as intelligent beeps.

I recommend watching this, please watch the whole 10 minutes if you want to give feedback on it.

https://www.youtube.com/watch?v=TqUJdDTo7v8

[Pete]

Pete, what would be the minimum specification for a machine to be self-aware? In principle, could a BBC Micro or ZX Spectrum be capable of self-awareness in the same way you believe more modern machines are?

Yes I think it would be possible to make a self aware conscious BBC micro. I have fairly recently bought a BBC Micro with a recapped PSU (if you find a BBC Micro the PSU must be recapped before first switch on otherwise the PSU can blow up). I'm sure the program would work well on the BBC Micro, the only thing I'd need to advance natural language processing and generation because the BBC Micro can't do the reaction to music as it is existing, but yes no reason why a 2MHz processor wouldn't be self aware, just as much as a 2GHz one.