AI driven expertise & profiling: hype, hope or déjà vu?

May was a busy month. Apart from helping establish then launch a real estate and mortgage business (Bees Homes) I was in Lisboa for Social Now and London for KM Legal UK.

I attended both in the expectation of learning more about the onrush of Artificial Intelligence and its implications for the Knowledge Management profession.

Specifically, I wanted to see how the encouragingly styled Talent and Knowledge Matching / Profiling systems might tackle the challenges of knowledge loss when people depart, of onboarding when people arrive and identifying / ranking expertise that might otherwise be opaque when pulling together teams.

It’s not a new topic: back in the late 90’s I was Business & Strategy Advisor to Sopheon PLC when we acquired Organik (a technology for identifying expertise) and built systems for US Insurers looking to establish the best teams for clients based upon expertise. We never cracked it even though we knew what the issues were (usually motivation)!

Seeking answers at SocialNow Lisboa while Keynote speaker Ellen Trude watches.

Armed with a list of ‘use cases’ I’d worked on with Martin White I set off in search of answers to these questions from both vendors and KM practitioners?

  • Onboarding: A new employee with many years of highly relevant experience joins the firm. How long will it be before their experience is ranked at the same level as their predecessors?
  • Legal: Is the profiling process compatible with the provisions of the General Data Protection Regulation? The thoughts of the Information Commissioner on this are worth a look. Profiling & Automated Decision Making
  • Functionality: Do they offer the ability to present a list of people ranked by expertise?
  • Language: In multinational companies where it is especially difficult to know all the experts, how does the vendor coppe with the fact that documents, meetings and social media traffic will be in local languages?
  • Chinese Walls: How does the application cope with expertise gained on projects that are secure, a common issue in law, finance and R&D where walls need to be erected to prevent commercial information being divulged>
  • Testing: What User Testing is undertaken with a client before signing a contract to verify that the profiling system works?

So, what did I discover? Thierry de Bailllon in his closing Keynote put it very succinctly but with a caveat:

Embrace or die? 88% of technologies already include AI.

Self reinforcing bias?

it’s not Enterprise Social Networks (ESN)!

This Twitter exchange between Ana Neves and Luis Suarez prompted by a question I posed of the Workplace (Facebook at Work) team following their presentation is revealing:

May 12 there’s been a few questions about expertise location 2017 I don’t remember that being the case in previous years #SocialNow

May 12 Well, I think people are starting to understand how critical it is to know who is who within the org beyond just content, right?

Replying to totally! It surprises me it took so long. It’s amazing the role #ESN can have in unveiling that expertise #SocialNow

On the surface the case for ESN is compelling. Yet the majority of vendors at SocialNow focus on information exchange and conversation rather than the capturing and cataloguing of it. One,@mangoappsinc, had a neat tool (they won the “coolest app” prize) with the ability to upgrade comments from threaded discussions and posts to create ranked knowledge resources from the mass of information and conversation.

So, ESN can show who has answered what question, conduct searches across conversations and in many cases act as a project management tool, the new Facebook at Work (Workplace) now allows the creation of documents for example.

Provided the application is linked to HR systems it is possible to retrieve profiles and see what expertise an individual might have. As one vendor (@OrangeTrail showcasing Facebook at Work)) who uses bots to generate responses put it:

‘Questions’ is the key to find experts as people don’t keep profiles updated.

I concur and they are great facilitation platforms though with advanced features that will suffice for many. Yet I left Lisboa though feeling organisations will need to rely on assisted search for some time if they want to take a deep dive into expertise

know what you don’t know

Peer Assist “Problems” for discussion

So onto London and KM Legal UK. An interesting Day One ended with a psuedo Peer Assist in which AI was raised a lot.

One observation (facilitation tip): the session failed to commit the ‘owner’ of the problem to action so as a result the feedback loop to plenary became a series of “we said this.”

Again, as in previous years I felt the focus was on operational tools and techniques which means that KIM Professionals in Legal are more at risk from the onrush of technology.

It reminded me of the issue Librarians faced with the arrival of end user search in the mid 90’s which finished their monopoly of being the people who found stuff in organisations.

Day Two took a deeper dive into technology and its potential impact.

AI in Legal today

This slide sets out where AI is making a difference in Legal.

I tweeted having heard Cliff Fluet’s excellent presentation:

Paralegals beware. AI is coming. Adapt or die?

And I questioned:

How wide is scope of AI? More than Doc Analysis / Creation. Opportunity to broaden knowledge base

As yet no one had focused on expertise and profiling so when one presenter cited the case where a newly arrived CEO asked the Head of HR / Talent Management to let him have profiles / competencies of the staff using their system it got my attention.

I asked whether the results the HR head gave the CEO inferred a level of expertise. It didn’t which got thinking that if the data set is incomplete and the issue of self reinforcing bias is not addressed then over reliance on one source for identifying ‘experts’ is dangerous. Imagine your career prospects if for whatever reason your name wasn’t on the ‘expert’ list given to the CEO?

and finally

So where do I see the state of expertise and profiling systems? Patchy!

Yes there are certainly companies who ‘get it’ but can they do it?

I am indebted here to Martin White who in an excellent report “People and expertise seeking – an overview” summarises the predicament thus:

The most important lesson learned is the need for an expertise location strategy that is linked into HR processes, knowledge management, training, job appraisals and social media development. Finding people with expertise is not a ‘search problem’.  Good search tools can certainly help but without attention being paid to profile quality (even if other types of content are being searched) and a commitment by employees to share their knowledge expertise discovery will not be as successful as anticipated or required.

My takeaways:

  • KIM professionals need a clear strategy (working in partnership with other stakeholders such as HR and IT) and be clear on the questions being solved by any system;
  • They need to be clear what they are getting, what’s missing and how it mitigates the potential for self reinforcing bias when they enter discussions with vendors around automating expertise seeking and profiling;
  • They need to recognise the importance of their role in facilitating the adoption of such systems and accept this is just a part of a portfolio of approaches of identifying, capturing and retaining expertise;
  • They need to be clear what critical knowledge actually is in their organisation and who is likely to have it in order to assess the veracity of the results of any pilot;
  • It doesn’t matter what solution you adopt, if your environment is not conducive to the sharing of expertise and people don’t see the value in it then save the money; and
  • In any event you cannot capture everything people know; we learn and share through stories (failures rather than successes) and those often remain hidden.

Practicing what you preach: when search is not enough for a Knowledge Base

Working independently you often have to solve complicated IT issues rather than calling the IT Help Desk as I did for 25 years when I worked in the City of London. 
A few years back resolving technical IT issues was nigh on impossible without retaining expensive external help. Today with networks, online search and a friendly local services business, its doable albeit time consuming. It does though allow you to experience the trials and tribulations associated with establishing and maintaining an effective Knowledge Base.

The context:

A decade ago I converted from being a PC user to a Mac user.  While I loved the order and structure of Windows Explorer navigation I found that Apple had cracked the user interface and synergy of devices. And it seemed to be less prone to Trojan Viruses.
So once Office for Mac appeared and I could use the tools most clients are familiar with I went the whole hog and over the next few years acquired a MacBook Pro, iMac and iPhone all of which sync seamlessly and have been fairly robust (up until now).
Having a deep distrust of relying purely on search it took me a while to create a navigational structure akin to Microsoft’s Explorer using Apple’s Finder tool but I got there mimicking what I had before on a PC.
Like many I used the folder structure in my Mail server to catalogue and group correspondence.  Mail was configured to mirror the file navigational structure so I thought I was ready for most eventualities..
Evernote and DropBox are the collaboration tools I use for current projects and where everything associated with them is stored. iCloud holds the most important historical contact data, Keychain the passwords and Time Machine which I back up to an external Hard Drive is the safety net for the Knowledge Base I’ve acquired over the past 20 years.

The issue:

A month ago my iMac’s Finder stopped working which meant I had to rely solely on Apple’s search tool Spotlight to find stuff on my Knowledge Base.  Despite a trawl thru various technological sites for a fix no one seemed to know how to get Finder working. Within a day I realised how dependent i’d become on ‘assisted search’

The advice:

Back up the iMac using Time Machine. Wipe it clean. Install the latest operating system (El Capitan) Export all material from the Time Machine back up but create a new profile rather than use the same profile. Should prevent the Finder problem being exported!!!!! All sounded good until I tried to export the data from Time Machine.

The subsidiary Issue:

I created a new profile on the iMac and was able to import most of the data but the majority of files can only be accessed thru my previous user profile which had different permissions. And Time Machine hides Mail folders (doesn’t show in Users>Name>Library>). So I couldn’t export the mailboxes as I’d originally intended to.

I felt I was spiralling down into the depths of uncertainty.

The final solution:

Find another machine, reinstall from Time Machine in my former profile and then save all the data (email/files/photos/videos) to an external hard drive. 

Sounded simple. In practice it wasn’t.
I had to borrow an iMac from the very supportive Fermin Ayucar at BeValued, recreate my system on that, save that using Time Machine to a hard drive and then wipe clear my own machine again before using the ‘new’ system to reinstall all of my Knowledge Base on my iMac.

And finally:

The sense of satisfaction when you ‘crack it’ and manage to rebuild a machine is immense. The sense of frustration though in not being able to locate stuff you know is saved is enormous and reflects what many people find.
 
It reinforces the piece I wrote a few month’s back on the need for assisted search and tags which contained this prophetic quote from my good friend Martin White of Intranet Focus who said:
 
If you can’t find information, then in effect it does not exist. Your search application may return 85,340 results for a query, but if the most relevant information was not indexed, or your security permissions inadvertently prevented the information from being displayed — can you trust your search application?
 
Fortunately the reinstall workaround was successful and all machines are backed and functioning. If you want more please contact me and I’d be happy to talk you thru it.

Fired up but not yet ready to go: Legal KIM response to 2015 challenges

A month back Martin White and I ran a breakfast breakout event for professionals in Legal Knowledge & Information Management. Those who follow mine and Martin’s musings might recall the event ‘The Future for Legal KIM: an outside in perspective’. Our aim was to present our thoughts to a group of Legal KIM’ers and seek their views.

These were the topics we foresaw as being important in 2015:

  • Lawyers come and go – capturing knowledge at speed
  • Collaboration and KM beyond the firewall
  • Getting the best from virtual teams
  • Bringing it all together – legal project management

ALegal KIMs it turned out we were not far off the mark as the feedback from the postcards we invited the delegates to write on indicated.

Once we consolidated all the replies on the day an interesting picture emerged.Scores on the doors (Click on the picture below to make it more visible)

In law firms of more than 250 partners the biggest issues were around virtual teams and project management. Yet all acknowledged they were not yet in a state of readiness to tackle them. Among the smaller players the biggest worry was around loss of knowledge.

other priorities

Not unsurprisingly the comments provided a valuable insight into their thinking.  Smaller firms (at the start of their KIM journey) were looking for basic KM:

  • Basic entry level km – completely new to it / evolving information research service & integrating with K activities
  • Provision of rapid and easy access to previously captured knowledge / Technology to simplify the process of intergration

Larger firms wanted something different (note the reptition of collaboration):

  • Collaboration inside the firewall / Expertise locating
  • Combining & improving KM systems / Organising our know how in a better way across the whole organisation
  • Support dept personalities working together (Marketing collaborating with IS, KM)  / Improve collaboration generally
  • Content clutter and records management / Risk & Security / Knowledge & UT goals & Strategy

So much to ponder on – watch this space for answers!

and finally

Grateful thanks to the four people who made contributions to Plan Zheroes (the event’s nominated charity).  For those who forgot and anyone else who feels moved to contribute, they can do so here.

PZ Virtual PresenceThis Thursday the Plan Zheores team are at London’s GLA for the launch of their new virtual presence which has the potential to make PZ the ‘Uber of surplus food’. Here’s a snapshot of what it will look like and why the team is so excited.