Perspectives #KMers #art-of-KM #KM-research


Sreejith Menath
 

The problem of scaling AI within an enterprise is not necessarily a technology problem rather its a knowledge management problem"  Dr.Harrick Vin, AISummit

My inclination towards Harrick's observation was further reinforced after going through illuminating perspectives of leaders in enterprise transformation like Raj Ramesh and Stan Garfield and decision analytics influencers like Cassie kozyrkov. We all have heard about technology and business innovation pioneers, who have either by exercising paradigm-shifting innovative force or by capitalizing smartly on a disruptive act of nature itself ( for example the new technological landscape emerging from a global crisis) , emerged as agents of change and leaders of that neo-movement resulting in a transformed profile of existing business reality. However, as tempting it is to believe that this effect is a very short term advantage, history offers standalone proof of the fact that those who sustain this advantage indeed have mastered the two most prominent anchors of differentiation when it comes to 'knock-out game' in disruption emergence.

These are principally belonging to two major lead fronts:

  • Assetizing by actively re-inventing enterprise 'people&process '( rather than separated entities, as a single aligned and communicative unit) capabilities.Your advantage here lies in smartly capitalizing on those models which exploit new trends and business landscapes, and integrating the two catalysts above in a unique way.This should be the guiding momentum behind the various cross-functional continuous improvement campaigns.
  • Creating 'differential knowledge artifacts' which can be continuously learned from data driven enterprise processes. When I say differential its not literally in the analytical or mathematical sense. Rather, what I imply is crystallizing knowledge in varied granularity that can be learnt, adapted, unlearned and re-formulated.

As far as the extended collaborative enterprise function is concerned, a federated learning based data alliance platform, is a promising avenue towards approaching these challenges strategically, in spite of some of the challenges that it brings along.

 In the past few months of my exploration in the knowledge management business, following the market leading products and trends, I got to realize a few challenges in this area, exposing certain gaps in their representation in enterprise knowledge management.

These are mainly as follows:

  • Major fraction of current market leading products focus on data re-allocation/rearrangement and personalization .The output thus created is mainly restrictive to the domain of semantic content intelligence or search intelligence.
  • There is an increasing focus on data quality and integrity, but the AI enabled SaaS market is still vastly focusing the advantage of the improved data availability on visualizations. Meaning that they show you the metadata constructs which underlie multiple heterogeneous data streams as well as qualitatively label them within a 'datahub' middle layer in data management architecture, but there is still lack of clarity on how to make meaningful use of those data dashboards.
  • Decision makers look for actionable intelligence, which is still mostly missing from the data dashboards. For example how the various KPI's relate and do they converge or diverge from the strategic organizational goals.

Content/search intelligence offers your organization a useful approach to "partitioning and consolidation" of numerous heterogeneous knowledge artifacts, that get scattered over in the process of conducting enterprise interactions. However, is content intelligence effective in driving the focus towards "meaningfulness and explainability" in how information assumes various instances in its lifecycle inside and across enterprise boundaries?

An organization should have a well laid out process to people (and vice-versa) maturity before it attempts any form of digitization based transformation. It has to undergo knowledge management capability assessment, that could derive solid guidelines on the missing links in adapting and evolving to new mold in enterprise information architecture.

We have data and we have the tools to "passively transform" it to a different format but from an end-to-end impact perspective do they serve the actual purpose of enabling a competitive advantage? Here are a few pointers that the decision-makers could possibly reflect and expand on-

To paraphrase, my central thesis is the following:

Analytics and business intelligence should sooner or later confront a shift from reporters of spatio-temporal status of enterprise data to manageable and system-aware pathfinders in the complex information occupying real-time integrated business continuum

Would be interesting to know the perspectives of active practitioners from the AI-enabled digitized enterprise community, on the proposals presented above.

As always optimistically looking ahead, to provoke disruptive thought-leadership that matters for real sustainable change


 
Edited

Easier said than done. 

The successful integration and exploitation of AI/BI/Analytics into an enterprise depends on exactly the same dynamics as any past technology advance. Those dynamics may be summarized by a simple principle I identified many years ago during my client consulting work:
You cannot automate that which you do not do (well) manually. 
If you take AI/etc. to be forms of automation, then this principle requires one to understand:
1. Which tasks, activities or decision processes is AI intending to automate?
2. Who is responsible for doing those things currently? (People)
3. How are those things done currently? (Process)
4. How well are those things done currently? (Quality)
5. What metrics are used to gauge how well they are done? (Metrics)

If the above cannot be described well and fully, then there is nothing to automate, and attempts to roll out AI/BI/Analytics/Visualization/etc etc will be disjointed and will meet with hit or miss adoption and success.

When we apply technology to a well-understood process that is already meeting with success, then we have a good chance of introducing automation to that process in a way that results in improving the outcomes. 

When considering AI/BI-type solutions, one would do well to consider the research on executive decision making, which confirms a high level of intuition being used, sometimes running contrary to the direction indicated by available data and evidence. Behavior-based norms like this can be difficult to overcome, making the introduction of decision support tools like AI challenging.

--
-Tom
--

Tom Short Consulting
TSC
+1 415 300 7457

All of my previous SIKM Posts


Brett Patron
 

Mr Short: 
You posted: The successful integration and exploitation of AI/BI/Analytics into an enterprise depends on exactly the same dynamics as any past technology advance. Those dynamics may be summarized by a simple principle:
You cannot automate that which you do not do (well) manually. 
 I would offer that you can still automate it. But it will not be any better than if you did not automate it at all. 

We have a saying on our team that if you "automate bad process you still suck, just faster."

 The "better living through technology" crowd will always try to sell the next cool toy. And the new toy is often more the Easy Button than a real solution. 

Unfortunately as we know as KMs, that makes our life painful over and over again.

Brett Patron, CKM
Deployable KM Specialist
US DOD Joint Enabling Capabilities Command
Norfolk, Virginia
@kmfordecision [Twitter]

On Tue, Aug 25, 2020, 10:48 AM Tom Short <tshortconsulting@...> wrote:
Easier said than done. 

The successful integration and exploitation of AI/BI/Analytics into an enterprise depends on exactly the same dynamics as any past technology advance. Those dynamics may be summarized by a simple principle:
You cannot automate that which you do not do (well) manually. 
If you take AI/etc. to be forms of automation, then this principle requires one to understand:
1. Which tasks, activities or decision processes is AI intending to automate?
2. Who is responsible for doing those things currently? (People)
3. How are those things done currently? (Process)
4. How well are those things done currently? (Quality)
5. What metrics are used to gauge how well they are done? (Metrics)

If the above cannot be described well and fully, then there is nothing to automate, and attempts to roll out AI/BI/Analytics/Visualization/etc etc will be disjointed and will meet with hit or miss adoption and success.

When we apply technology to a well-understood process that is already meeting with success, then we have a good chance of introducing automation to that process in a way that results in improving the outcomes. 

When considering AI/BI-type solutions, one would do well to consider the research on executive decision making, which confirms a high level of intuition being used, sometimes running contrary to the direction indicated by available data and evidence. Behavior-based norms like this can be difficult to overcome, making the introduction of decision support tools like AI challenging.

--
-Tom
--

Tom Short Consulting
TSC
+1 415 300 7457

All of my previous SIKM Posts


 

@Brett wrote:

I would offer that you can still automate it. But it will not be any better than if you did not automate it at all. 
We have a saying on our team that if you "automate bad process you still suck, just faster."
😆


Fair point! (I’ve heard it said another way: “automating the outhouse.”). 
--
-Tom
--

Tom Short Consulting
TSC
+1 415 300 7457

All of my previous SIKM Posts


 

 PS - ergo the inclusion of (well)!!
--
-Tom
--

Tom Short Consulting
TSC
+1 415 300 7457

All of my previous SIKM Posts


Murray Jennex
 

well said Tom, this is the basis for most systems analysis and design courses.  Of course you can automate directly something you don't do well (i.e. not fix the process) but then you tend to make the list of bad examples that we love to share in these classes....murray


-----Original Message-----
From: Tom Short <tshortconsulting@...>
To: main@SIKM.groups.io
Sent: Tue, Aug 25, 2020 7:48 am
Subject: Re: [SIKM] Perspectives #kmresearch #artofkm #kmers

Easier said than done. 

The successful integration and exploitation of AI/BI/Analytics into an enterprise depends on exactly the same dynamics as any past technology advance. Those dynamics may be summarized by a simple principle:
You cannot automate that which you do not do (well) manually. 
If you take AI/etc. to be forms of automation, then this principle requires one to understand:
1. Which tasks, activities or decision processes is AI intending to automate?
2. Who is responsible for doing those things currently? (People)
3. How are those things done currently? (Process)
4. How well are those things done currently? (Quality)
5. What metrics are used to gauge how well they are done? (Metrics)

If the above cannot be described well and fully, then there is nothing to automate, and attempts to roll out AI/BI/Analytics/Visualization/etc etc will be disjointed and will meet with hit or miss adoption and success.

When we apply technology to a well-understood process that is already meeting with success, then we have a good chance of introducing automation to that process in a way that results in improving the outcomes. 

When considering AI/BI-type solutions, one would do well to consider the research on executive decision making, which confirms a high level of intuition being used, sometimes running contrary to the direction indicated by available data and evidence. Behavior-based norms like this can be difficult to overcome, making the introduction of decision support tools like AI challenging.

--
-Tom
--
Tom Short Consulting
TSC
+1 415 300 7457

All of my previous SIKM Posts


Ryan Fitzgerald
 

To consider another perspective, I have automated tasks and created a solution for a process that didn’t exist without automation. The automation made the task feasible, the process would not have been undertaken without the automation in the first place.

 

Cheers,

Ryan.

 

From: main@SIKM.groups.io <main@SIKM.groups.io> On Behalf Of Murray Jennex via groups.io
Sent: Tuesday, August 25, 2020 1:19 PM
To: tshortconsulting@...; main@SIKM.groups.io
Subject: [EXT] Re: [SIKM] Perspectives #kmresearch #artofkm #kmers

 

well said Tom, this is the basis for most systems analysis and design courses.  Of course you can automate directly something you don't do well (i.e. not fix the process) but then you tend to make the list of bad examples that we love to share in these classes....murray

-----Original Message-----
From: Tom Short <tshortconsulting@...>
To: main@SIKM.groups.io
Sent: Tue, Aug 25, 2020 7:48 am
Subject: Re: [SIKM] Perspectives #kmresearch #artofkm #kmers

Easier said than done. 

The successful integration and exploitation of AI/BI/Analytics into an enterprise depends on exactly the same dynamics as any past technology advance. Those dynamics may be summarized by a simple principle:

You cannot automate that which you do not do (well) manually. 

If you take AI/etc. to be forms of automation, then this principle requires one to understand:
1. Which tasks, activities or decision processes is AI intending to automate?
2. Who is responsible for doing those things currently? (People)
3. How are those things done currently? (Process)
4. How well are those things done currently? (Quality)
5. What metrics are used to gauge how well they are done? (Metrics)

If the above cannot be described well and fully, then there is nothing to automate, and attempts to roll out AI/BI/Analytics/Visualization/etc etc will be disjointed and will meet with hit or miss adoption and success.

When we apply technology to a well-understood process that is already meeting with success, then we have a good chance of introducing automation to that process in a way that results in improving the outcomes. 

When considering AI/BI-type solutions, one would do well to consider the research on executive decision making, which confirms a high level of intuition being used, sometimes running contrary to the direction indicated by available data and evidence. Behavior-based norms like this can be difficult to overcome, making the introduction of decision support tools like AI challenging.

--
-Tom
--

Tom Short Consulting
TSC
+1 415 300 7457

All of my previous SIKM Posts

Confidentiality Warning:

Deloitte refers to a Deloitte member firm, one of its related entities, or Deloitte Touche Tohmatsu Limited (“DTTL”). Each Deloitte member firm is a separate legal entity and a member of DTTL. DTTL does not provide services to clients. Please see www.deloitte.com/about to learn more.

This message and any attachments are intended only for the use of the intended recipient(s), are confidential, and may be privileged. If you are not the intended recipient, you are hereby notified that any review, retransmission, conversion to hard copy, copying, circulation or other use of this message and any attachments is strictly prohibited. If you are not the intended recipient, please notify the sender immediately by return e-mail, and delete this message and any attachments from your system. Thank You.

If you do not wish to receive future commercial electronic messages from Deloitte, forward this email to unsubscribe@...

Avertissement de confidentialité:

Deloitte désigne un cabinet membre de Deloitte, une de ses entités liées ou Deloitte Touche Tohmatsu Limited (DTTL). Chaque cabinet membre de Deloitte constitue une entité juridique distincte et est membre de DTTL. DTTL n’offre aucun service aux clients. Pour en apprendre davantage, voir www.deloitte.com/ca/apropos.

Ce message, ainsi que toutes ses pièces jointes, est destiné exclusivement au(x) destinataire(s) prévu(s), est confidentiel et peut contenir des renseignements privilégiés. Si vous n’êtes pas le destinataire prévu de ce message, nous vous avisons par la présente que la modification, la retransmission, la conversion en format papier, la reproduction, la diffusion ou toute autre utilisation de ce message et de ses pièces jointes sont strictement interdites. Si vous n’êtes pas le destinataire prévu, veuillez en aviser immédiatement l’expéditeur en répondant à ce courriel et supprimez ce message et toutes ses pièces jointes de votre système. Merci.

Si vous ne voulez pas recevoir d’autres messages électroniques commerciaux de Deloitte à l’avenir, veuillez envoyer ce courriel à l’adresse unsubscribe@...


Murray Jennex
 

I think your last bullet about decision makers and actionable intelligence is the key bullet.  AI is still a technology that finds patterns in data.  KM is the discipline that can assist organizations in differentiating those patterns that are meaningful and the limits of the patterns applicability.  a major risk/threat of AI is over generalizing the applicability of an identified pattern and I think it is up to KM to capture the knowledge from the pattern as well as understanding why the pattern works so that reasonable limits can be applied to the application of the pattern knowledge....murray


-----Original Message-----
From: Sreejith Menath <sreejithmenath1985@...>
To: main@SIKM.groups.io
Sent: Tue, Aug 25, 2020 2:27 am
Subject: [SIKM] Perspectives #kmresearch #artofkm #kmers

The problem of scaling AI within an enterprise is not necessarily a technology problem rather its a knowledge management problem"  Dr.Harrick Vin, AISummit
My inclination towards Harrick's observation was further reinforced after going through illuminating perspectives of leaders in enterprise transformation like Raj Ramesh and Stan Garfield and decision analytics influencers like Cassie kozyrkov. We all have heard about technology and business innovation pioneers, who have either by exercising paradigm-shifting innovative force or by capitalizing smartly on a disruptive act of nature itself ( for example the new technological landscape emerging from a global crisis) , emerged as agents of change and leaders of that neo-movement resulting in a transformed profile of existing business reality. However, as tempting it is to believe that this effect is a very short term advantage, history offers standalone proof of the fact that those who sustain this advantage indeed have mastered the two most prominent anchors of differentiation when it comes to 'knock-out game' in disruption emergence.
These are principally belonging to two major lead fronts:
  • Assetizing by actively re-inventing enterprise 'people&process '( rather than separated entities, as a single aligned and communicative unit) capabilities.Your advantage here lies in smartly capitalizing on those models which exploit new trends and business landscapes, and integrating the two catalysts above in a unique way.This should be the guiding momentum behind the various cross-functional continuous improvement campaigns.
  • Creating 'differential knowledge artifacts' which can be continuously learned from data driven enterprise processes. When I say differential its not literally in the analytical or mathematical sense. Rather, what I imply is crystallizing knowledge in varied granularity that can be learnt, adapted, unlearned and re-formulated.
As far as the extended collaborative enterprise function is concerned, a federated learning based data alliance platform, is a promising avenue towards approaching these challenges strategically, in spite of some of the challenges that it brings along.
 In the past few months of my exploration in the knowledge management business, following the market leading products and trends, I got to realize a few challenges in this area, exposing certain gaps in their representation in enterprise knowledge management.
These are mainly as follows:
  • Major fraction of current market leading products focus on data re-allocation/rearrangement and personalization .The output thus created is mainly restrictive to the domain of semantic content intelligence or search intelligence.
  • There is an increasing focus on data quality and integrity, but the AI enabled SaaS market is still vastly focusing the advantage of the improved data availability on visualizations. Meaning that they show you the metadata constructs which underlie multiple heterogeneous data streams as well as qualitatively label them within a 'datahub' middle layer in data management architecture, but there is still lack of clarity on how to make meaningful use of those data dashboards.
  • Decision makers look for actionable intelligence, which is still mostly missing from the data dashboards. For example how the various KPI's relate and do they converge or diverge from the strategic organizational goals.
Content/search intelligence offers your organization a useful approach to "partitioning and consolidation" of numerous heterogeneous knowledge artifacts, that get scattered over in the process of conducting enterprise interactions. However, is content intelligence effective in driving the focus towards "meaningfulness and explainability" in how information assumes various instances in its lifecycle inside and across enterprise boundaries?
An organization should have a well laid out process to people (and vice-versa) maturity before it attempts any form of digitization based transformation. It has to undergo knowledge management capability assessment, that could derive solid guidelines on the missing links in adapting and evolving to new mold in enterprise information architecture.
We have data and we have the tools to "passively transform" it to a different format but from an end-to-end impact perspective do they serve the actual purpose of enabling a competitive advantage? Here are a few pointers that the decision-makers could possibly reflect and expand on-
To paraphrase, my central thesis is the following:
Analytics and business intelligence should sooner or later confront a shift from reporters of spatio-temporal status of enterprise data to manageable and system-aware pathfinders in the complex information occupying real-time integrated business continuum
Would be interesting to know the perspectives of active practitioners from the AI-enabled digitized enterprise community, on the proposals presented above.
As always optimistically looking ahead, to provoke disruptive thought-leadership that matters for real sustainable change


Murray Jennex
 

yes Ryan, and I agree that you had to understand the process well to do this and that you couldn't do the process manually goes with what we are saying, if automating a process take advantage of the technology to make it better and don't just automate what is there.....murray


-----Original Message-----
From: Ryan Fitzgerald via groups.io <ryafitzgerald@...>
To: main@SIKM.groups.io <main@SIKM.groups.io>; tshortconsulting@... <tshortconsulting@...>
Sent: Tue, Aug 25, 2020 10:30 am
Subject: Re: [SIKM] Perspectives #kmresearch #artofkm #kmers

To consider another perspective, I have automated tasks and created a solution for a process that didn’t exist without automation. The automation made the task feasible, the process would not have been undertaken without the automation in the first place.
 
Cheers,
Ryan.
 
From: main@SIKM.groups.io <main@SIKM.groups.io> On Behalf Of Murray Jennex via groups.io
Sent: Tuesday, August 25, 2020 1:19 PM
To: tshortconsulting@...; main@SIKM.groups.io
Subject: [EXT] Re: [SIKM] Perspectives #kmresearch #artofkm #kmers
 
well said Tom, this is the basis for most systems analysis and design courses.  Of course you can automate directly something you don't do well (i.e. not fix the process) but then you tend to make the list of bad examples that we love to share in these classes....murray

-----Original Message-----
From: Tom Short <tshortconsulting@...>
To: main@SIKM.groups.io
Sent: Tue, Aug 25, 2020 7:48 am
Subject: Re: [SIKM] Perspectives #kmresearch #artofkm #kmers
Easier said than done. 

The successful integration and exploitation of AI/BI/Analytics into an enterprise depends on exactly the same dynamics as any past technology advance. Those dynamics may be summarized by a simple principle:
You cannot automate that which you do not do (well) manually. 
If you take AI/etc. to be forms of automation, then this principle requires one to understand:
1. Which tasks, activities or decision processes is AI intending to automate?
2. Who is responsible for doing those things currently? (People)
3. How are those things done currently? (Process)
4. How well are those things done currently? (Quality)
5. What metrics are used to gauge how well they are done? (Metrics)

If the above cannot be described well and fully, then there is nothing to automate, and attempts to roll out AI/BI/Analytics/Visualization/etc etc will be disjointed and will meet with hit or miss adoption and success.

When we apply technology to a well-understood process that is already meeting with success, then we have a good chance of introducing automation to that process in a way that results in improving the outcomes. 

When considering AI/BI-type solutions, one would do well to consider the research on executive decision making, which confirms a high level of intuition being used, sometimes running contrary to the direction indicated by available data and evidence. Behavior-based norms like this can be difficult to overcome, making the introduction of decision support tools like AI challenging.

--
-Tom
--
Tom Short Consulting
TSC
+1 415 300 7457

All of my previous SIKM Posts

Confidentiality Warning:
Deloitte refers to a Deloitte member firm, one of its related entities, or Deloitte Touche Tohmatsu Limited (“DTTL”). Each Deloitte member firm is a separate legal entity and a member of DTTL. DTTL does not provide services to clients. Please see www.deloitte.com/about to learn more.
This message and any attachments are intended only for the use of the intended recipient(s), are confidential, and may be privileged. If you are not the intended recipient, you are hereby notified that any review, retransmission, conversion to hard copy, copying, circulation or other use of this message and any attachments is strictly prohibited. If you are not the intended recipient, please notify the sender immediately by return e-mail, and delete this message and any attachments from your system. Thank You.
If you do not wish to receive future commercial electronic messages from Deloitte, forward this email to unsubscribe@...
Avertissement de confidentialité:
Deloitte désigne un cabinet membre de Deloitte, une de ses entités liées ou Deloitte Touche Tohmatsu Limited (DTTL). Chaque cabinet membre de Deloitte constitue une entité juridique distincte et est membre de DTTL. DTTL n’offre aucun service aux clients. Pour en apprendre davantage, voir www.deloitte.com/ca/apropos.
Ce message, ainsi que toutes ses pièces jointes, est destiné exclusivement au(x) destinataire(s) prévu(s), est confidentiel et peut contenir des renseignements privilégiés. Si vous n’êtes pas le destinataire prévu de ce message, nous vous avisons par la présente que la modification, la retransmission, la conversion en format papier, la reproduction, la diffusion ou toute autre utilisation de ce message et de ses pièces jointes sont strictement interdites. Si vous n’êtes pas le destinataire prévu, veuillez en aviser immédiatement l’expéditeur en répondant à ce courriel et supprimez ce message et toutes ses pièces jointes de votre système. Merci.
Si vous ne voulez pas recevoir d’autres messages électroniques commerciaux de Deloitte à l’avenir, veuillez envoyer ce courriel à l’adresse unsubscribe@...


Sreejith Menath
 
Edited

Relevant remarks there!.Just to clarify,I was not in particular focusing on the sole imperative of automation or without it.I was just indicating towards the gaps! and why they invite attention and how are the ways to deal with them.
However good to know the other angles.