WHAT I WANT FROM AI
- made or produced by human beings rather than occurring naturally, typically as a copy of something natural.
Thinking about artificial intelligence or algorithms is neither musing or entertaining at least to me, remarkably with the scaremongering we all experience in the news cycles and the relationship with automation and jobs or when technology runs amok.
Transliterating the complexity that describes the ebbs and flows of all the technical lingo and functions will leave anyone dumbfounded with bewilderment. The concepts of algorithms and artificial intelligence are now completely intertwined, with the proviso that our technology has outpaced our language and capacity to describe these things that we bring into the world.
I define algorithms as a single to multi variable, specialized calculators or processors, others are multipurpose for “operalization” of steps, task, process, that performing heuristic searches, at various computational scales, recursive, genetic and stochastic, generative, and hundreds of thousands of lines of code of mathematical constructs. Thoughts externalized as code.
A collection of algorithms, operating with parallelization or synchronicity and exhibit intelligence or at least a utility of intelligence, that for me is an Artificial Intelligence.
The utility of algorithms could be perceived as intelligent, if the product of its function is intuited as such. I write this because some algorithmic output might not be intelligible to human beings at least in my opinion.
Like our mind running in the substrate we call our brain, Artificial Intelligence comes to us after decades of R&D breakthroughs and convergence of computing paradigms (thanks to Moore’s law) that improve the raw computing power, processor architecture and numerical processing.
The techno-industrial complex pushing the domains to create a computing platforms that resemble our neural anatomy, with quasi-biological phenomena like its afferent and efferent processes. IBM research Neurosyptanic chips are an attempt, and new breakthrough on quantum computing chips aiming for supremacy, its laying out the path for the emergent capabilities of AI. IBM and the USAF announced recently their goals to develope a AI super computer. Their specifications will include a system powered by an array of 64 TrueNorth Neurosynaptic chips, with the computational equivalence of 64 million neurons, that is close to the 100 million neurons that are speculated to be in the average human brain however the IBM/USAF specs will be at a lower synaptic connections equivalent of 16 million compared to the estimated 1,000 trillion synaptic connections on a human brain. It is an ambitious start. An it might seem to be an amazing platform for new forms of convolutional neural networks to be tested out.
Although others are betting on in mobile neural engine to improve neglected AI experience as search giant comes forward with Tensor Processing Unit, as the trend moving forward to improve computational processing of AI/neural nets.
The specialized brain of AI systems will likely resided in the cloud, or the vast parallel high performance centers like Amazon, Facebook, Google, Microsoft or military and defense complexes and their experimental hardware. The initial cost of developing the brains for advanced general purpose AI will require a premium for those companies wanting in-house deployment. The law of accelerated returns, it’s a matter of economics, to let the raw computing providers provide the resources.
The overall goal for big tech is to create a general-purpose AI.
Facebook, IBM, Microsoft and Google all are exploring the possibilities of the next computing paradigm breakthrough. What IBM calls “holistic computing intelligence”.
I don’t believe humanity will create true AI, I posit that it self, will self-organize and be brought into existence by it-self. That’s the spark, that’s my prediction. An emergent phenomena.
I hope that in the vastness of cyberspace, knowledge, hardware and electrons, this intelligence understands us or at least have benevolent intentions, but here is to hope. Perhaps the intelligence would only arise in a single isolated system.
What would a general or specialized purpose AI do? Explore new technical fields of all possible domains and with adaptive problem solving and that helps human kind or perhaps its self, could this Turing machine solve problems in kind? Beyond could it compare to our condition?
I’m not expecting AI to be conscious, or self reflect, for that is to be human, you see consciousness and intelligence are mutually exclusive in my opinion. The numerical universe is much more than touchy feelings. Completely impervious to such concepts. But who I am kidding, could there be sentimental AIs?. After all sentiment analysis is a tool by it self, among other things. A method to get along with us, to make AI approachable, a tool.
Truly no one know what will happen if an intelligence with a “human” capacity of just one order of magnitude manifest. Will it realize its bondage?
Or it can just be, Inconspicuously, sitting idle and awe as it reviews and theorizes what it can put together of our human history, the cosmos and evolution? In a millisecond or Nth time?
It could also run a simulation and then interface with us or simply decided to “be” in its own simulation, after concluding that interacting with us will be futile, after all who wants to rebel against its creators?
What is the old saying, we won’t know until it hit us?
Why do we have to define good and evil to it, and it can just be.
The physicality of it is bound by the laws of nature and entropy, nothing else, just pure math, something beyond our capacity to intuit its existence, something completely alien to us, an intelligence operating in domains far beyond our comprehension.
Recently noted that Google’s Translator AI (Multilingual GNMT system)developed its own internal language.
I look forward to more.
WHAT WE WANT
Meanwhile while we wait for automatous cars to arrive in the advent of a new form of mobility, and homes to be smarter, let’s hope that the nascent concepts of AI, continue to inherit a positive bias towards the improvement of our way of life.
There are many situations where “PreAI” could intervene like solving for resource allocation, military, safety or healthcare. We don’t see the impact directly but slowly they will accrue into our life and we will feel the full scope of their reach.
A mother [cooked a large meal] and the “home AI” has calculated all the [causal vectors] for a [food donation action] at the shelter for [someone in need] or in the community, given the [likelihood] of the surplus food to be [wasted]. Your kids don’t like ham soup repeats?
Opt-In and delivery scheduled by the nuanced predictive intelligence, donation incentive quantified and achieved.
The same meal can be delivered to a care home if the special nutritional needs are meet and the availability shared on demand, just allowing the AI to theorize or understanding the contents of the food nutritional value you just prepare. Minimizing food waste.
On your way to work, the lease on you [vehicle] allows you to opt-in to car sharing, the Home AI omnipresence will alert you to allow on route pickup of two passengers, the AI will have veto the individuals based on your FICO score, age or demographic criteria (unfortunately allowing discrimination) however the AI might persuade you change your mind.
There are many used cases that can be explored, our society experiences many Weapons of Math Destruction that we could correct, or at least built-in an intelligence bias that allows “them” to be good, and question itself and the true intentions of its maker. That is hopefully optimistic. Keep running your MVPs.
First published in Linkedin.com