Are PDFs fit for purpose?
Are PDF's fit for purpose in today's complex world.
The Portable Document Format (PDF) was created by Adobe Systems in?January 1993 - more than 31 years ago. And it has been an extremely successful standard for the exchange of documents.
Today we are bombarded with 100's of documents: reports, policy, strategy, roadmaps, submission and recommendations - all available as PDF's. I checked the Analysis & Policy Observatory (APO) which archives such documents and found that last year there were 38 documents related to Defence . And in my area of interest (AI, Robotics, Manufacturing and Innovation) there 100's more. And worse still - many of these documents reference each other (as they should).
For example, there were 79 Submission to the National Robotics Strategy.
I just don't have the time to read (and analyse) them all.
And I am retired - imagine how much less time any executive has.
So we are now resorting to using AI to summarise these documents .
But this is telling us something.
We need to question - What is the purpose of these documents?
For the most part, these documents are to help us make decisions.
They help us make decisions about our investment in our future.
And if we need AI to help us work out the meaning of a document.
Then - PDFs are clearly not fit for this purpose!
Decisions are best supported by the appropriate Knowledge Management (KM).
"Knowledge management (KM) is the collection of methods relating to creating, sharing, using and managing the knowledge and information of an organization."
There are many Knowledge Management technologies .
In theory, the best solution would be to put all the knowledge into one database - called a Knowledge Base (KB). I suspect that this was the intention of GovTeams.
But the reality of human nature (and government departments) is that everyone lives in their own silos - where they can control access to their own knowledge.
Of course, the biggest KB is the WWW - but this is unstructured data.
A potential solution has been Linked Data and the Semantic Web.
Whilst this technologies has some success, such as life sciences, cultural heritage, and government open data initiatives, it is not pervasive.
领英推荐
Of course now we have chatGPT that can make sense of the WWW.
Except that - the WWW does not make sense - there is a great deal of non-sense.
Try asking chatGPT "What is the defence procurement policy of Australia?"
And it will just give you a list of PDFs.
So what is the solution?
I have been playing around with Obsidian - article on Linkedin and Obsidian
Obsidian is a Personal Knowledge Management (PKM) which stores all of your knowledge in a vault - which is a folder on your file system which contains your notes, attachments, and the configuration settings. It is able to quickly create links from one note (idea) to another, along with tags and file properties. It is able to show backlinks (very useful) and a visual graph of your knowledge. It comes with some very sophisticated search functions and a vast array of plugins. There are now plugins that allows AI to search your own PKM. Personally I have found this to be an amazing tool for understanding the relationship of complex ideas.
So, what I was thinking - could we use Obsidian (or similar PKM) to create a Portable Knowledge Management?
Rather than a department or organisation writing a PDF document, they spend their time creating a customised vault that contains the information relevant to topic at hand. Once finished the vault is zipped up and published (a .pkm file)
This file can be downloaded (or emailed) and automatically unzipped.
From the users point of view - the experience would be the same as clicking on a PDF file, except rather than firing up a PDF viewer, it would fire up a PKM viewer, such as Obsidian. From here the user can navigate the Knowledge Base - conducting searches, and potentially asking questions with AI.
This is just a random thought - but imagine if it were possible to merge PKMs.
In theory, PKMs could replace many existing PDFs - like user manuals.
Which leads to the image of a robot reading a roadmap.
Why is this so anachronistic?
Because no ones uses roadmaps anymore - we all use a navigation app.
The app tells us to how to navigate from point A to point B.
So why can't we do the same for decisions that affect our future.
Why can't we have an app that helps us navigate the nation from A to B
Emeritus Professor at University of Queensland
8 个月Very true. We are going to get there I think if this ill-advised hoopla about AGI ends and people start focussing on more modest yet very useful applications of LLMs.
?? 中国广告创新国际顾问 - 综合数字传播客座教授 - 140 多个创意奖项 ?????
8 个月Seems like a bothersome PDF pursuit! Any dynamic tools to sail through data overload? ??
Great insights on the limitations of using PDFs for submissions. A well-designed Knowledge Management system would definitely streamline the process!
Non-Executive Director, Chair and Advisor in mining, metals, manufacturing & industry research
8 个月Nice logic Elliot… at what point will we just need to corral information, not even assemble it, and point an AI bot at it to deliver useful interpretation for someone’s final decision? And is your final thought an example of AI replacing politicians, or just guiding them to navigate from A to B where B is the best outcome for Australia?
Independent Robotics Research Consultant
8 个月I just found the Data and Digital Government Strategy https://www.dataanddigital.gov.au/resources And notice their strategy is on line (there is still a PDF). But there does not appear to be a Digital Strategy for creating and sharing Strategies. I recall back in 90s, CSIRO's policy on creating policies had expired - and they could not legally create policies. They had to get an extension from Parliament. Very Yes Minister.