Centralized AI - The Mind in the Middle

Written by: Mark Cowan - CEO, Put It Forward

May 30, 2019

Do little bits of smart make a lot of smart?  Leveraging AI in a practical sense.

Everyday we hear about technology solutions that are incorporating this AI feature or that Machine Learning (ML) capability all in the name of making things better.

Looked at a different way - does localized AI actually have a material benefit on the whole?

In this post we’re going to look at the benefit and the peril of having applications with embedded AI capabilities.

Technology vendors are tripping over themselves to differentiate by offering powerful capabilities leveraging AI technologies.  Including ML and natural language processing (NLP).

The implication of localized intelligence within a process or an application means it will have a positive impact on the whole.   From the design perspective what do we reference to get the model right? There is only one model that we can look to for reference clues - nature.

The Reference Design

Throughout nature (human beings included) - intelligence is centralized within the organisms brain with a nervous system radiating outwards.  (There are exceptions with invertebrates like starfish and sea urchins etc. which do not have brains and leeches which multiple brain segments).

The benefit of a centralized nervous system (CNS) are many but at its base it’s the shortest path between two extreme points thus allowing signals to move as efficiently as possible while being coordinated.

Additionally the brain has its most important receptors very close if not directly attached to it.  The sensors being sight, sound, hearing, touch, smell and cognition.

The Augmented Design

There really is no reference design for making a piece of something smarter and how it will interact with other parts of the system.

Let’s consider it in the context of the reference design - if you were suddenly give the ability to have one smarter part of your body for instance a hand or a kidney how would you perform overall?

So if it’s one of your hands - does that become the better hand - no, it still takes its commands from the brain.  Why - because if it didn’t, you’d have an interoperable system. So if this is obvious why would it be any different than what happens when AI is introduced into a system?  The answer is, it’s likely not.

I Robot - TM & Copyright © 20th Century Fox Film Corp

In the above picture from the movie I Robot - The three characters represent a natural CNS, a hybrid and a synthetic CNS.  The hybrid gains marginal advantage in limited scenarios.  Although this is the movies.

The implications being there is marginal benefit of AI being introduced into an overall system unless it contains a design strategy for interoperating with everything else.  If it does not - then the benefit is localized at best. At worst it doesn’t offer a sustained improvement as it’s dependant upon the signal data from the other parts of the system and likely a power drain overall.  Meaning more energy will have to go into the overall system to keep it running.

The Centralized Closed Loop Design

Leveraging the natural design is the starting reference point

Again we look to the operational functions of marketing, finance, customer, hr and service for the working example.  Mostly because they have a tremendous amount of signal data and transactional uniqueness to them.

For it to function three things need to operate in harmony to close the loop when executing

  1. Operational data - signal data for telemetry
  2. Intelligent analysis - insight and analytics
  3. Orchestration - when and where events

What we really mean by utilizing the benefits of AI capabilities is that it has to benefit everything, not just the part it’s attached - else we are still in the paradigm as only being as good as the weakest or slowest part.

For the operational functions of marketing, finance, customer, hr and service this means being able to hyper-personalize each interaction, segment your audiences intelligently not just basic demographics, link the behavior of prospects and customers in the fastest way possible and have deep forecasting insights.

We’ve got a lot to say and offer on this approach, we believe it’s critical to get it right the first time as this is when the organizations of today will determine if they’re around for the next wave of change.

Related articles
The Fundamental Guide to the Modern Customer Data Platform The customer data platform market is worth over $2.4 billion. It's a popular form of business software across almost every...

Continue reading

6 Ways Marketers Can Use Machine Learning 6 Ways Marketers Can Use Machine Learning in 2019 We already know that artificial intelligence and machine learning technology will shift...

Continue reading

7 Steps to AI Closed Loop Marketing - Infographic Sometimes the best blog entry is an infographic, more pictures - less words is a good thing right?  We've broken down all...

Continue reading

Complete A/B Testing Reports With Analytics Often people want to get a deeper understanding of email A/B testing results than what's available in standard marketing analytics. Generally...

Continue reading

How is AI Empowering Sentiment Analysis? Sentiment analysis is a new area of technology that's being looked at as the next natural step in digital communication. To understand what...

Continue reading

Seeing Through the Click - Content Awareness Seeing Through the Click - What’s Happening Inside the Content Often, our customers and prospects ask for some insights into how their...

Continue reading

Everything You Need to Know About Snowflake Data Integration Snowflake's data cloud solution has already been adopted by over 4,000 organizations. It's one of the best data platforms...

Continue reading

Experience Continuity - Hyper Personalization Through Linked Experiences Experience Continuity - Hyper Personalization Through Linked Experiences When Walt Disney first created...

Continue reading

How Householding Customer Data Unlocks Personalization At Scale Problem: Generalized messaging across customer groups where no obvious linkages exist is because segmentation is...

Continue reading

How to Create a Customer Experience Strategy: Case Study Today's customers demand better experiences from the brands they interact with. To improve your customer experience, you...

Continue reading

How to Unlock Snowflake Super Powers With Put It Forward It may be an understatement that quality decisions are based entirely on the data quality going into the Snowflake data...

Continue reading

Identity Resolution - Customer Experience Core Infrastructure Leveraging What You Have Most companies already have everything that’s needed for identity resolution. It’s not about...

Continue reading

Financial Planning and Analysis Automated One of the most critical business processes in any company is financial planning and analysis (FP&A) because it provides leadership...

Continue reading

The Fundamental Guide to Account-Based Marketing CMI's Enterprise Content Marketing research shows that 34% of businesses use account-based marketing and 21% plan to incorporate...

Continue reading

The Sitecore Water Cooler Podcast - How to get signals from data Organizations and enterprises today are presented with enormous opportunities. However, they’re also presented with...

Continue reading

10 Tips on Improving Your Big Data Analytics Are you looking to improve the amount and the reliability of your big data analytics outcome? Are you unsure of what aspects you should...

Continue reading

Vendor Roadmaps:  What You Need To Know Written by:  Mark Cowan, CDO, Put It Forward 22 January 2019. The more information you can get about vendor intentions, the better, right?...

Continue reading

What Is a Data Platform? In an enterprise, seeing and storing data is not enough; understanding it is the key to extracting insights that can increase business value. A data platform...

Continue reading