The biggest thing in tech the last few weeks has been GPT-3. GPT, or Generative Pre-trained Transformer, is a language model developed by OpenAI that uses deep learning to produce human-like text. It is based on the Transformer architecture, which was introduced by Google in 2017. GPT uses a self-attention mechanism to learn the context of a given sentence and generate text that is both coherent and natural sounding. It has been used to generate text for tasks such as summarization, dialog generation, machine translation, and more. GPT is a powerful tool for natural language processing and is one of the most widely used language models today.
What if I told you that the previous paragraph was not written by me, but by a machine? That “wow!” moment you just had is one of many to come, as we realize how powerful artificial intelligence (AI) is becoming. We are witnessing feats performed by machines where we can no longer easily tell the difference between human and computer.
What makes GPT-3 so good?
To start, it is the sheer size. GPT-3 has 175 billion trainable parameters in its model. The next closest model has 17B. A training run uses 45 TB of data, almost 10 days of run time and over $10M in compute power. To give you some perspective, the bot you are using on your job posting site probably trains in less than 3 mins. GPT-3 then takes this sophisticated model and huge training dataset and mixes it with the key ingredient: human expertise. In three waves, humans provide feedback to the machine, help it learn, and reinforce good outcomes. Yes, AI is not magic. Humans play a critical role. Version 3 of this model took 3 years of engineering, training, feedback, and revisions.
Can business use GPT-3?
GPT-3 can be broken down into two primary functions: language understanding and generative AI (composing answers). With the former, GPT-3 excels at interpreting human language and deriving meaning from it. With the latter, it creates its own answers, but those responses can be wildly incorrect, making it a less-than-ideal choice for enterprise use. Additionally, GPT-3 is built on public information that is two years old, so it cannot handle current events, making it especially ill-suited for enterprise use. Let’s compare a Google search to GPT-3.
Google’s real-time results
ChatGPT results based on 2021 information
In a business setting when someone asks about how much vacation time they get, the answer depends on their employment status, their union contract and even the country they reside in. GPT-3 can’t handle that type of personalization. Further, you want your organization’s bot to understand your lingo, your acronyms, and your slang. GPT-3 is one model trained for everyone. Its training data is not your organization’s data. Even though GPT-3 has limits in an enterprise setting, it will no doubt influence the future direction of IT departments.
How will GPT-3 influence enterprise software?
Because our blog readers are most interested in enterprise software, we must examine the effect GPT-3 will have on the enterprise. To start, it will change user expectations. Let’s take a trip down memory lane….it was 2008 and Steve Jobs had just launched the App Store followed by the famous ad campaign, there’s an app for that! I remember this year well. Most of our customers never spent a moment thinking about mobile or mobile apps. After all, enterprise software was too complicated for that tiny screen!
Apple’s Famous Ad Campaign: There’s an App For That
We can sometimes forget that our enterprise users are consumers, too. They go to movies, buy groceries, go to restaurants, seek entertainment, and so on. Enterprise users, like everyone else, fell in love with mobile apps, came back to work or school with new expectations, and started demanding mobile apps for enterprise functions. Schools and organizations scrambled. Everyone wanted an app. There were RFPs every day calling for proposals from vendors.
Do we think it will be any different this time? ChatGPT (the chatbot built on the GPT-3 model) will work its way into consumer life. Your users will relish the productivity gains it provides. Think about how we find and retrieve information today. We use Google and Google gives us pages and pages of links that all look the same. We have no idea where to click. If we ask that same question to GPT, we get a two-sentence answer that is right on the money. Imagine that instead of pointing, clicking, searching, reading, we just asked and got the answer…
The future of user interface is language
Students and employees are routinely on the search for information; be it about a policy, a form to fill out, a piece of data or analytics. Now imagine a user interface powered by chat AI that gives you an answer quickly, succinctly, and with accuracy that beats human beings! The old way of searching intranets or knowledge base articles is going the way of the dinosaur. (Or should I say punch card?) The flaws of the search result page UI are already being exposed.
GPT will also prove the demand for one-stop shopping: One place to ask any question. While each of these SaaS applications may have some semblance of a bot, they all live on islands without any connection to each other; islands with no bridges. ChatGPT has shown users the value of being able to handle any question you throw at it.
Coming Soon: Ida and GPT
GPT-3, while not perfect, certainly has some capabilities that are pushing the AI universe ahead by enormous leaps. Ida, IntraSee’s very own digital assistant, operates much like GPT-3, but at a different scale, where we build and train to each client’s uniqueness. Starting with version 23.01 due in April 2023, Ida will introduce its first integration with GPT-3. We can’t tell you much about it now, but we will tap into the power of the GPT-3 model to improve upon the secured, personalized answers Ida provides. More to come with the release notes of 23.01.
The next release of Ida, our digital assistant, is now available. Clients can talk to their account teams about a deployment schedule that works for you.
22.04 in Summary
Ida 22.04 releases many housecleaning items and general fixes. The focus of new features started with improved support for a new ODA Web Channel complete with chat transcript storage, voice channel support, autosuggest, improved read more functionality and much more.
Next we have many new reports available and new/improved intents and PeopleSoft Campus data sources for Exams, Deadlines, Course Catalog and Registration Appointments.
Finally, we are adding an advanced Find/Replace tool for fixing answers in one swift click and the ability to schedule bot down time to bring the entire bot down and present the user a message about its status and when it will be available again.
Release Notes
Automated scheduled downtime for entire bot for support of maintenance windows
Configurable list of allowed languages for translation
New Class Deadlines Intent
New configuration options for Help Escalation
New Course Catalog Search intent
New dedicated config pages for adapters
New Find/Replace Answer Text Tool
New frequent Users report
New Ida bundled Live Chat integration
New Rater Progress Report(s)
New Registration Appointments Intent
New Suggestions reports
New Question Confidence Regression Report
Salesforce adapter ticket logging enhancements
Better automated duplicate feedback handling in High Training/Value modes
Fixed an issue where FBL was showing the wrong sub-org’s answer text
Fixed issue with long topic labels in KPI tile
Improved Dashboard Conversation Log
Improved disambiguation help support
Improved error messages for unexpected remote answers for Salesforce
Improved FAQ Read More styling
Revised Exam Schedule Intent
Streamlined client user IUC console access permissions
User counts in reports now more closely reflect distinct guest users
Web channels: Replaced lightboxes with Slide Out support
Web channels: removed debug web console messages
Web channels: dynamic down message support
Contact us below to learn more and setup your own personal demo
Despite the current buzz, artificial intelligence (AI) isn’t magic. AI is just a collection of probabilities, but those probabilities feed on data and human input. Yes, humans play a critical role in AI. For example, the newly hyped GPT-3 model has 3 phases of human input. Alexa and Siri have teams of humans helping the AI grow “smarter.” In short, your AI can’t perform without the human AI experts. But what if you don’t have a team that lives and breathes this technology every day?
Introducing Oracle Digital Assistant Tune-Up
Your digital assistant is a digital worker. Just like your human workers, they need an annual review, training and feedback. IntraSee is now offering our annual Oracle Digital Assistant (ODA) Tune Up service to give your digital worker that annual review. The ODA Tune-Up will engage our highly trained AI teams to review your current digital assistant approach and training data to eliminate misunderstanding and increase accuracy in your bot. The Tune Up service will provide actionable feedback using our empirical methods and AI testing tools to show you exactly where your bot may need optimization.
IntraSee has been building and running Oracle Digital Assistants (ODA) since the product was released over three years ago. As one of the first companies to deploy these bots, we’ve learned quite a bit. To achieve success in your chatbot project, accuracy is job #1. We’ve spent the past three years figuring out just the right methods for maximizing bot accuracy on ODA.
Our clients are consistently achieving greater than 90% accuracy from their NLP (natural language processing) as a result. For your Machine Learning application to achieve high accuracy, you need a strong model and well-tuned training data; and to get those, you need AI experts. At IntraSee we use a team of data scientists, AI architects and computational linguists to produce our best-in-class accuracy results.
Schedule Your Bot “Check Up” Today
Users tell us the most common reason they won’t use a bot is because they don’t believe it can help. In other words, the bot didn’t understand their questions. With improved understanding, adoption will increase, greater service will be provided, and more ROI unlocked. Just as humans see a doctor each year for a check-up, your bot needs an accuracy “check-up” from the leading team of experts in this field.
For a limited time, we are offering an attractive introductory rate. To learn more, please contact us below.
The next release of Ida, our digital assistant, is now available. Clients can talk to their account teams about a deployment schedule that works for you.
22.03 in Summary
Ida 22.03 focuses on continued automation, improvements to non-authenticated chat analytics and UX improvements to translated conversations along with the routine bug fixes and other minor enhancements. Our goal is to fully automate the training and deployment of Ida bots and 22.03 moves us a couple steps forward.
Additionally, this release introduces beta support for the Oracle SDK web channel UI. This web channel offers a different look and feel as well as some new features. The first is the ability to voice chat over the web with Ida. Ida can understand your voice and also respond with its own voice. This new channel also supports persistent chat logs as you move from page to page and auto-suggest questions to improve accuracy from its already stellar performance.
Release Notes
Entity event handler support
Streamlined UI Kit javascript library used in Ida web channel
Beta support for ODA SDK ChatUI, including: voice support, type ahead, saved conversation summary and more
Improved bot training automation (beta)
Improved top topic chart in IUC console
Sub-org rating suggestions now filter out peer orgs
Feedback loop now lists rating history of who have rated/approved/rejected
Fixed an issue with top questions report in Federated Mode
FAQ upload supports additional answer providers
Improved same intent dialog while in DA context
Improved conversation tracking for guest users
Support for fuzzy entity matching
Various federated mode fixes
Consolidated conversation creation logic
Maintenance mode for content/data providers (take systems offline with graceful bot responses)
Resolved an issue with duplicate rows in feedback loop
New dedicated config pages for integration adapters
Improved FBL display for utterances sent to help
Fixed an issue where interactions were not showing in Feedback Loop
Corrected behavior with feedback loop with legacy data
Export/import now available for all training data
Safeguard prevention of FAQ use of certain question types (Frustration, Greeting, Help)
Pre-translation text warning users the response is auto-translated
Fixed an issue with Thumbs ratings on non-content based FAQs
Additional safeguards to training data in production
Stopped enforcing “list max check” for FAQ Question Inputs
UI support for FAQ question overrides
Contact us below to learn more and setup your own personal demo
Gartner recently released their Hype Cycle for 2022. The Hype Cycle is a theory around the 5 phases of emerging technologies. Chatbots fell into Phase 3, the Trough of Disillusionment, this year. While that title sounds scary, it is a pivotal moment for an emerging technology where the best will be separated from the rest. In this post, we will review the chatbot market, where it was, where we are now and what is coming next.
Gartner 2022 Hype Cycle
Where we were
Around 5 years ago Natural Language Processing (NLP) took a big leap. We had a breakthrough in artificial intelligence (AI) and Machine Learning techniques where our ability to have machines understand human language took a jump. Suddenly, chatbots, as they were dubbed, were all the rage. This irrational exuberance, perhaps fueled by Alexa, created a flood of demand. Rest assured; supply always finds a way to meet demand. At this point, everyone and their sister built a chatbot.
Many of these bots were rushed. Some used fake AI and their NLP accuracy was abysmal to match. The experience felt more like AOL Keywords than it did a machine who truly understood the human. Bots were released covering many singular use cases from FAQs, to recruiting, to CRM systems and even enterprise applications like HCM. The investment ranges were so wide you could find a bot for $10k a year or build one on IBM Watson for $1M.
This rush just served to flood the market with confusion. Customers began looking at price as the most distinguishing factor. Suddenly, we were all apples and, boy, were there a lot of apples. How did we like them apples? Well, not so much…
Where are we now
With the market flooded with poor quality and confusing messages, customers have become disillusioned. Was this all hype? Will it ever work? This feels just like IVR (Interactive Voice Response)!
To add to the confusion, sales teams were proclaiming AI magic. It just learns! It is automatic! If your success is less than automatic, you are going to be disappointed. You may even write the technology off (though we would caution you to look again).
That AI breakthrough we spoke about earlier? Well, that happened when we learned (pun intended) to mimic how the human brain works. The human brain divides everything into buckets like apples, oranges, bananas and so on. When we see a new fruit never before seen, we just know it is a fruit. But this only works with lots of learning. To learn, we read, we attend class, we discuss, we experience. Bots learn by consuming data and instructions from data scientists.
So, while the cheap bots fell way short of our expectations, the enterprise bots seemed complicated to implement and manage. Organizations threw people without AI experience into AI projects.
Reality then punched us in the face. And so the disillusionment phase has arrived. Let’s dive into where today’s chatbot projects have failed. If we can learn from this phase, we can be better for it. We can focus with more depth on what makes a magical conversational experience.
Why chatbot projects fail
Many chatbot projects failed because many customers were caught at one of two extremes – both driven by the desire to minimize cost. On one side, the poor-quality bot which required little effort yet never delivered on the salesperson’s, “It’s magic!” promise. On the other extreme, customers put in significant effort to try and build their own bot only to realize building and running AI at scale is not the same skillset they historically have possessed.
False No-Effort Narratives
“No-Effort” implementations mean you are getting a generic product; Too many customers have been sold this and failed. Your users see right through a generic bot which really is no better than classic IVR.
While there is such a thing as AI which learns on its own, it is a major liability for your organization. Facebook and Microsoft have failed spectacularly in this regard.
Some customers attempt pilots to prove out the technology, but pilots rarely work. As an organization you have to be all-in on making automation work. Further, these are solutions that evolve over time and a short pilot is no time at all.
Crawlers and Links
Real AI personalized to your users and organization is hard. Some projects attempt to shortcut this hard work by having the bot “crawl” in existing web site for content. Remember that salesperson? It’s automatic!
Crawlers are how search engine’s work and if your name isn’t Google, those searches have been shown to not perform well nor do they have any personalization.
Our focus groups have shown that users have no faith in search not named Google. Disguising search as chat will ensure a lack of adoption.
Per Gartner CIO surveys, digital transformation is nothing without personalization.
Bots which do not know you, where you do not log in, just end up sounding like that IVR.
Every question can have dozens of answers solely based on the attributes of the user. Most bots today fail to personalize, and the results show.
No Breadth
Many of the early bots simply focused on a small sliver of questions, like recruiting or financial aid. As we talk about in Higher Ed Chatbot Usage: Perception vs. Reality, your users don’t care about your reasons to only tackle a sliver.
Our data shows users will ask every question that comes to mind, including questions completely unrelated to the page they’re on.
If your bot cannot handle at least 250 intents, you stand little chance of success.
Comfort of Live Chat
Live-chatting features sound like a good idea. I can always talk to a human! Keep in mind, you need to staff/train/manage that human.
Live Chat is not the killer feature organizations should be requiring. AI accuracy is. Live chat just covers up the failure to do job #1.
Even more, vendors know Live Chat is a parachute, albeit one that eats ROI significantly. Knowing they have a parachute, they rest easy and don’t solve the hard problems.
Let’s Jump into a use case related to this topic. Candidate.ly recently published some survey data on recruiting bots. This is just a peak into a particular vertical and what the customers thought of their chatbots.
Candidate.ly Survey on Chatbot Effectiveness
You will notice some 71% thought their bots were average or below. If the parent thinks their child is average, then you are probably dealing with a D student. It is no wonder why organizations are disillusioned, but it doesn’t have to be this way.
According to the report, “Two in five people (42%) avoid chatbots when making a complex inquiry and 15% lack confidence in using technology to contact organisations”. And according to the Institute, if technologies like chatbots were “well designed and implemented”, most customers would be happy to use them.
This one report highlights what we all have experienced. We want chatbots to help us, but many of them are just falling short. Our expectations have shifted downward. However, this is the moment you can surprise and delight your users if we really focus on designing the experience.
…if technologies like chatbots were “well designed and implemented”, most customers would be happy to use them.
What comes next: Enlightenment
We started with irrational exuberance and now find ourselves in the Age of Disillusionment. Disillusionment just means that customers are no longer buying into the hype without a critical eye. They are now in prove-it mode. The success of the early adopters who chose wisely will begin to show. Those who chose poorly will cautiously wade back in, but will be much better informed. Bad solutions and methodologies will fade away and the market will gain clarity. We will be on a steady incline that is more predictable and sustainable.
We welcome this phase where the market will gain clarity and users will have their needs met. It is often lost that chatbots are really automation projects. You are replacing a function normally performed by a human. That means your best chance of success is to focus on outperforming the human. You have to be more accurate, more consistent, more available and more accessible including speaking your language.
When you pick a bot, remember that cheap is expensive when it comes to automation. If your bot does not offer any advantage when compared to speaking to a human, then user behavior won’t change. If behavior doesn’t change, you won’t be able to reduce spend and then you are simply paying for both the bot and the human.
…cheap is expensive when it comes to automation
Good results take effort. Good results require personalization and NLP accuracy. The bot is mimicking a human and just like humans, they need to learn and grow. You are feeding them and should treat the bot like an evolution and not like a one-time project.
If you are ready to ramp up the bots, here are the top strategies we recommend you follow on your project:
Pick a platform/product that you can influence and that will learn about your organization. Be sure it is based on Machine Learning AI.
Commit to a three-year initial Agile cycle. Implement, monitor, learn and evolve in increments no longer than 3 months. Learn from your users!
Prioritize breadth and personalization. This is the key to user adoption.
Create a funnel to maximize ROI. Drive users to the automated path first and let the bot pursue escalation paths such as live chat when needed. Do this after the initial learning period of six months.
Make the bot omnipresent. The bot should meet users where they are, not just on one website. That chat icon should be everywhere.
If you have any questions or want to see a demo of Ida, our own bot, please reach out below. As an industry, we are approaching the beginning of enlightenment; we believe our clients are already there. We hope that if you have experienced a failed chatbot project, you will try again with fresh eyes. If you need any help or just want to talk things through, just drop us an email! 😊