There were plenty of search engines before Google came along and grabbed a lot of the search market share. There were plenty of generative AI solutions before ChatGPT appeared and created a large amount of buzz too. The question is why did they grab attention / users and what can we learn from that to apply to OSS solutions?
This image below from Aviva Pinchas compares Google and Yahoo from circa 2000. There’s a significant difference right?
Google provided a single search bar for the user to drive their own info-finding experience. Search and filter. By contrast, Yahoo provided categories and moderation as well as lots of possible choices.
Google also provided personalised, contextual and even iterative search (ie awareness of what a user searched for previously) within its algorithm. It attempts to consider searcher intent / context based on the question and previous uses. It also incorporates additional information such as current location, search history and more. This gives a more customised outcome for a market of one (ie each individual user).
Having previously been a Yahoo searcher, I made the switch (many years ago) to Google because it made it more efficient to find the info I was looking for. It made it easier to ask the question and get contextual answers. It reduced the amount of my limited cognitive capacities required to find what I needed.
But now, ChatGPT (a generative AI) has come along and changed the dynamic again. ChatGPT has provided a simple, conversational user interface that has made it more efficient to find the info I was looking for (depending on what info you’re seeking).
Let’s ask an OSS-centric question and compare the results:
Google’s response is:
Notice how the response from Google is cluttered and possibly distracting, forcing the user to make a decision from many different options. The response pane is not dissimilar to the Yahoo search console in terms of “information overload.” Google is great, but makes the user have to sift through various links in hope of finding the answers they’re seeking. The business model of selling ads means that paid links may also get in the way of finding the information you need.
By comparison, ChatGPT’s response is:
The ChatGPT user doesn’t have to go looking through various webpages to find the answer they’re looking for. ChatGPT is optimised for natural language. It’s also iterative, allowing users to refine, improve and extend the questions they ask.
If we compare the two, ChatGPT was the faster way of finding a coherent answer to the question.
To summarise, Google took the industry by storm by streamlining the experience of asking questions (and reducing cognitive load). ChatGPT took it a step further by streamlining the question AND answer experience (reducing cognitive load further).
Most OSS user interfaces still look like Yahoo’s search console, similar to the mock-up below.
What if instead we had a conversational question and answer interface to unlock information from all the amazing data our OSS and BSS collect? What if we could iterate and refine the questions / answers? What if we could ask it to show graphs, maps, forms, data, reports, text and more? What if we could incorporate additional information such as location into the mix for greater awareness of the user’s situation? What if it could provide attribution to show how the inference was derived.
Or, if we were to take our UI designs to the next level, what if we never even had to initiate engagement with our OSS? What if the solution would only bring us the information that was in most desperate need of our attention… and even gave us a recommended Next-Best Action (NBA)?
What if that situational awareness and NBA / decision-support was delivered to us via a head-up display (HUD) using Augmented Reality (AR) technology in the field?
Would it surprise you to know that none of this is science-fiction that will be possible many years from now?
There isn’t one single solution that integrates all of these pieces together today (that I’m aware of), but I have already seen every piece of the puzzle working in isolation. There are AR headsets that will arrive on the market soon that will completely revolutionise our current ways of working (in much the same way that smartphones did a decade ago). The software building-blocks are all available now (awaiting further iterative improvements of course).
The pieces will all come together quickly for the world of telco / OSS too once the right headset appears. If you’re not already experimenting with these next-generation solutions, then your competitors will have a significant head-start over you.
If you do want to embark on a journey towards next-gen user-interfaces but don’t know where to start, leave us a note and we’ll connect you with some innovative suppliers to potentially partner with.
Caveats:
We know there are limitations with ChatGPT. Accuracy of data (hallucinations), appropriate attributions, lack of ability to provide a range of choices to some questions (eg where to find the best price on a pair of shoes in my area) and more no doubt. Data governance, recommendations / actions and choices are all important factors for consideration as part of next-gen OSS solutions too, but this article is more about the UI designs and reduced friction / cognitive-load.
I can envisage using Google and ChatGPT (best of both worlds scenario), depending on the type of info being searched. Given the vast number of different personas that interact with OSS/BSS data and ways that they could consume that data, I can also envisage a horses-for-courses approach to UI design.
PS: An example of a conversational search overlay to an OSS could be the combination of this project (https://towardsdatascience.com/use-chatgpt-to-query-your-neo4j-database-78680a05ec2#d083-cf615c9d7f04) applying ChatGPT to Neo4j (which happens to be the graph database that underpins our sandpit inventory solution). Thanks to Seshan for the hat tip on this project! Sounds like something to play around with in my sandpit…. if not for being swamped with customer work at the moment. Another task gets added to the to-do list!! 🙂
This article by Joshua Yu also provides a working example of linking ChatGPT for creating graph data queries – https://medium.com/@yu-joshua/building-an-academic-knowledge-graph-with-openai-graph-database-12b320f08ef0
Have you ever had an OSS or BSS project that’s impacted by poor data quality? The problem of data integrity gets raised on almost every
“One of my favorite things about designing software is designing the defaults. The defaults define the experience for everyone out of the box. And, therefore,
Have you ever been tasked with mapping out the processes for the telco you work for? Either as a way of documenting current state, or
In the last week alone, I’ve had two potent reminders that the world of OSS is simply (pun intended) too complex. This manifests in many
Here at Passionate About OSS, we are exactly that – Passionate About OSS (Operational Support Systems). We’re also passionate about BSS, NMS, or any other names you call the tools that help to operationalise your network.
