Aime was born out of an idea of what ‘intelligence’ actually was. The idea was far too big to build all of it straight away – and so we started small.
Really small.
The first stage led us to create an intelligent (but fun) mobile phone app, which got to know you – and then pretended to be you (under your control of course!) A description of Aime Messaging can be found here, and information on how the app was used is here.
Aime Messaging was impressive, but it was essentially a toy. For our next product, we decided to make it commercially useful. Aime For Business (Aime4B) was a tweeked version of Aime Messaging which gave the user some business functionality.
The core on which the mobile products had been built was essentially still a type of AI – it seemed intelligent, but, like all AI, it didn’t understand anything. The next steps started to change that.
The next step saw the original core being used as a bootstrap process into the beginnings of an informational structure which started to try to understand the concept of everything it came across. This was tested as an automated chatbot production system (which could read your company website, learn what your company did and answer questions about it), but was never productised (maybe when we have a bit more time!)
What we really needed was the ability to be able to teach the AGI how to do things in the real world. We started off with the simplest possible process we could think of:
Add 1 to 0
That took six months! But progress was cumulative. A month later we had taught the AGI how to multiply – in such a way that it could work out how to divide on its own – and maybe more. A couple of months later we had something that we could start to teach to be a consultant, solving business problems across market verticals and continuing to learn on its own.
That is where we are now.
Next steps? You’ll just have to wait and see – but its going to be good.