Artificial Intelligence Singularity!

At the singularity point, the laws of physics don’t apply anymore. Dig deep into the black hold and reach out to the centre – there you have the singularity point.

We are in some ways in the AI Singularity point – at least for some type of prediction applications that are applied in the enterprise.

We are NOT talking about: 

• Image recognition AI – interpreting photographs, X-rays and merrily giving out predictions.
• Voice recognition AI – like Alexa – they continue to assist you in your lives.

We are specifically talking about AI used to predict the future based on past history – with specific reference to business numbers.

With COVID, AI algorithms are confused:

• Netflix’s algorithm must be confused. Now, the home bound customer is watching movies even on weekdays. The movies he is watching is not fitting into the earlier pattern.
• How can Amazon serve the best products to you if the markets leading products are masks and sanitizers? The algorithm had to be tuned.

 So, what will happen to business systems if they rely hugely on AI?

• The inventory ordering AI: Based on the March numbers, they will predict a massive back to school season in May-June and order inventory.
• The quarterly target setting AI: Will give meaningless numbers to the field members without factoring in the disastrous reality.
• The training recommendation AI: There is a huge upswing on training consumed by employees – the recommendation AI will struggle to be relevant as most of the training are consumed immediately.

So, we are at the AI Singularity Point – where the normal predictions of AI do not apply.

The Sheldon Plan

That brings us to Hari Sheldon’s Sheldon Plan. Recall your Asimov – the Foundation Series. Sheldon is the mathematician who develops a way to predict society way out in the future in mathematical terms.

Sheldon’s AI based pre-recorded message will appear during ‘Galaxy changing’ Sheldon crisis and guide the leader on the next steps. Over millennia it guides leaders to shorten the chaos after the fall of the galactic empire.

This is an AI that works even during crisis.

How do we have our business prediction systems work even during crisis?

What is the use of business prediction systems if it cannot predict the ebb and flow of the business? Or if it cannot give a leading indicator when the water is going to turn into an un-raft able Class 6 rapid?

Again, it is not the AI’s problem. It is the limitedness of the AI’s definition in the interconnected world.

Butterfly and the westerlies

Just like the flutter of a butterfly in China can affect the westerlies, the jump of a virus from one species to another can affect your store walk-in next week.  [Yeah, the original connects the flutter to a tornado].

How does the AI factor this? I have questions, no answers.

How do we get events external to the enterprise factored into the AI system? Will there be companies (like CB Insights) that provide AI Learning APIs for AI Systems of the future to factor in external events?

• A WW Economic Leading Indictor Feed that sends out metrics including purchase managers index, Brent Futures, Baltic Dry with regional feeds
• A Social Indicator feed that factors in relative peace and stability in the society (As of now on 1^st June 2020, for US it will indicate ‘BOILING’.
• A Pandemic Indicator feed that will factor in virus and other vectors that can affect business parameters
• A Conflict Indicator feed that will tell you how markets and supply lines will get affected if the conflict
• A ‘Sun Font Size’ feed if you may, to find out the relative importance of an incident to the common man. NYT, WSJ seem to be so boring and uniform. May be in the future, your normal run of the mill AI system will have daily newspaper syndicated feeds to understand the world around.
• You can even put in an Asteroid Indicator if you don’t want your business to get disnosaured! Asteroid indicators can come from VC investment data and can predict how your business can be demolished by new innovations.

So, in the AI system of the future could have:

• Proprietary company specific data feeds
• Environment Feeds whose ‘learnings’ are provided by third party providers through APIs as a paid service

to come up with reliable predictions.

Who knows how many circular references we will have in the feed:

• The lack of butterfly flutter in China may lead to a weaker westerly
• This might cause El Nino
• Which might lead to no rains in the Amazon basin
• Which might lead to lesser consumption in Brazil
• Which might lead to lesser demand in China’s factories
• Which might lead to more traditional hunting
• Leading to near absence of butterflies to flutter in China
• … And the whole cycle goes on.

It is like Harry Bellefontaine song, ‘There is a hole in the bucket Dear Lisa’.

Until such time business prediction AI engines do not factor in objective reality, yeah, there will be a hole in the bucket.