What AI is NOT

10 September 2017

(Reflections on AI by Invacio's Founder - William James Dalrymple West.)

First off, I despise writing or even explaining this, but having been involved directly for AI on my project for 5-ish years, I feel compelled to in-light of all the BS that is floating around.

As mentioned I’ve now been working 18–22 hours a day 6 days a week, for just shy of 5 years on my project Invacio, it’s a mammoth project, last count some 500 million lines of code without the 2 billion or so collected, derived or other datasets.

I am not some major corporation nor have I had investment outside of the scope of my own, and that of my family fund, it’s been an endeavor like nothing before in my professional career and one that leaves me both exhausted and ready for more.
During this development before AI was ‘it’ I had to build a team, 70 at its height, and we were working through numerous theories with no rule book drawn and are involved in all variations of AI.
If we take a moment to determine the main core ‘types/stages’, as I put them in regards to our smallest segment (financial markets).

Type I AI: Reactive machines (Archimedes)
The most basic types of AI system, are purely reactive, and have the ability neither to form memories nor to use past experiences to inform current decisions.

Type II AI: Limited memory (Agnes)
This Type II class contains machines can look into the past.

Type III AI: Theory of mind (Aquila)
We might stop here, and call this point the important divide between the machines we have and the machines we will build in the future.

Type IV AI: Self-awareness (Jean)
The final step of AI development is to build systems that can form representations about themselves.
We at Invacio are at for sure, based on all indicators, borderline stage IV with multitudes of outlets/inlets between or methods, as it’s best described.

AI is a tool, or is at this stage a severe advanced stage of computations, the AI that you read about in Musk or other comments is here but still very far off, it’s achievable, but we have so many things to learn about ourselves first before we can create full Sentient AI, I’d say True AI is a stage down from Sentient and where we are crossing over, Sentient is still quite some major investment and time away, which brings us to another point, [I neither care whether people agree or disagree, unless you are knees deep in building AI you really do not have a clue] it’s one thing employing people another actually being involved.

The human mind/body is perfect but imperfect, and this is without the ‘personality/conscious side’ that we still do not fully grasp or understand.

Until this is fully equated no matter the amount of money thrown at it, an AI True or Sentient will literally just be a mould of the people who created it, and not ‘True’ just a very advanced Bot, functionality, and capability wise.
So when you see such on the news, like I did for a man I admired, I am asking myself what his reason or what others reasons are, because when I look at what ‘all the large companies are bringing out IBM, Facebook, Google, Amazon etc., I am surprised they are trying (Musk/Other) to sell the fear.

We launched internally video analyzing for our social core, and our financial core a year before Google released theirs, and something else before Amazon theirs, essentially I feel the larger companies need to catch up somewhat, seeing Microsoft Glasses this week I was in horror simply using on — demand captured still via the glasses processing through their internal API and reading out in General Language a response for what was being observed.

This is watering down AI or showing a severe underlining lack of innovation or deliverance at these organisations, and then on the other hand you have (as mentioned above) Musk screaming the world is going to end.

No it’s going to be disruptive, so disruptive people do not realise, I speak to accountants, financiers, bankers, lawyers all the time, and I just sit there and in bewilderment over their un-lack of understanding of how insignificant they will be in a few years (or our case, now), like lambs to the slaughter.
AI in its current state is dynamic driven self-generating, learning algorithms and databases, not some Borg like creature, everything else is BS.

But that’s careers, jobs or other, that’s not death by machines, Christ, the machines we have now have a bad enough habit of walking without falling over, we’re years from getting them going properly then to add the AI element… much longer, and this is without the pure power to run, for Jean we run off 60 servers running at 10% clusters, and what not, and then the database servers, local servers and so on.

Without a single public user…..
Hardware will the ‘spade’ in this game, and a lot of oomph needs putting in for the power-requirements of AI just to run it, which is where immense costs come in.

So sure, change is coming, but not death by robots, just career death by algos, and to be frank why not, it’s a technical revolution and change is coming, but the point is we as Humans are pretty intense internally, there is copious amounts we need to learn about ourselves first before we can even consider to understand the development phase of full blown sentient AI, for now, we will settle for ‘True AI’ which I would ‘move’ to below Sentient AI as a separate stage.