The COVID crisis saw a huge rise in use cases and adoption of chatbots and AI, hammering home the benefits in crystal clear HD for businesses and governments working under pressure. As the restart gathers pace, adoption will continue as companies look to do more with less, and as the benefits of bots are more obvious.
Writing about bots for the last few years has been a painful affair. Typically, a brand launched a bot with a modest fanfare, there would be a smattering of stories and the bot would never be heard from again. LEGO, eBay, British Airways and many more all launched bots that vanished into the digital ether.
Worse, some organizations sneaked them out the side door as tests or trials. These vanished even faster because – funnily enough – people won’t use bots if they don’t know about them. Chatbots live in a funny hybrid state, companies wanted to test them quietly, but bots need a lot of noise, interactions and data to make people aware and for the bot to get smarter.
COVID has helped change all that. Companies needed overnight solutions to massive communications problems that affected national and local governments, healthcare and businesses.
Just one example, “Brian Pokorny had heard of AI systems for call centers before. But as the IT director of Otsego County, New York, he assumed he couldn’t afford them. Then the pandemic hit, and the state governor ordered a 50% reduction of all government staff, forcing Pokorny to cut most of his call center employees. Meanwhile, inbound calls were rising as more residents began seeking reliable COVID-related guidance and medical information.
So Pokorny picked up the first solution that landed on his desk. Within days of his signing up, the team helped him deploy a chatbot to address callers’ most common questions, such as how to identify symptoms or how to get tested. The software also made it easy for him to update and expand the chatbot’s responses as queries evolved.”
For many others, all those articles suddenly surfaced as they searched for help, many from 2017/18 in the hype years of bots when there wasn’t much evidence on show. Newer pieces can backup the hype with numbers. They advise business leaders, IT departments and communications executives that they can launch a chatbot in days (not months) that they could update fast-changing advice live and that feedback from bots could improve the service in minutes, not after months of A/B testing and focus groups.
And the recent numbers only help to reinforce and demonstrate the value of bots. IBM reported a 40% increase in traffic to Watson Assistant from February to April of this year. The World Health Organisation’s Health Alert service reached 12+ million people via WhatsApp, with total messaging through Facebook’s up by more than 50%.
Specialist vendors too saw huge growth in inquiries, demand and bots launching to help. SnatchBot’s independent COVID advice chatbot hit two million visitors within days and the use of bots developed on the company’s cloud platform has rocketed. I asked company founder and CTO Avi Ben Ezra whether this represented a sea-change in chabot use.
“It absolutely does. I’m happy to share our metrics. In February 2020, we were running at around 600 new registrations a week, with about eight hundred new chatbots being created a week. Now, mid-June 2020, over a thousand people a week sign up to build chatbots on our platform and 1,433 chatbots were constructed with us last week. In the past twenty-four hours, our chatbots have exchanged nearly 500,000 messages.
“If we can suppose other chatbot companies are experiencing similar demand, then it’s clear there is a massive surge of chatbot creation and deployment taking place right now across the globe.”
Of course, the successful bots from the likes of Dutch airline KLM’s Blue Bot, Bank of America’s Erica have long-guided their business thinking and they were better prepared to update their bots with COVID information. UK Bank Nationwide updated its Arti chatbot with COVID information for mortgage holidays and fielded 10,000 queries in a matter of days.
KLM are happy to discuss the growth of use from their bot, with some useful insights:
- “Since the outbreak of the coronavirus we are monitoring our social service volumes and the question types closely.
- As airspaces closed unexpectedly mid-March, our volumes grew tenfold.
- To make sure our service agents can focus on the most challenging cases, we automate frequently asked questions caused by the COVID-19 pandemic in our automation solutions. Examples of solutions: chatbots, automated service messages or other artificial intelligence (AI) integrations.
- Early on, our bots were able to answer English questions about ‘flight cancellations’ with more questions related to ‘flight status’, ‘vouchers’ or ‘general coronavirus questions’ added soon after.
- Thanks to the integration of language detection, we were able to add answers in Dutch, Spanish, German and Portuguese.
- Our latest integrated bot answer is related to the requirement to wear a face mask onboard. As the coronavirus is changing travel, passengers have a big need for information. AI is very helpful in addressing this.
- The share of handling service volumes by a bot hugely increased in March & April (from 8.1% to 22%).
- We can also support our service agents by, for example, already collecting data. Thanks to the latest developments, the percentage of cases fully handled (from 0.4% to 10.9%) or assisted (from 7.7% to 12.6%) by automation solutions massively increased in March & April.
- The higher these percentages are, the more time we save for our service agents to focus on the more difficult cases where personal assistance is really needed.
April saw Google launched its Rapid Response Virtual Agent, a special version of its Contact Center AI, and lowered the price of its service in response to client demand.
Of course, there are still plenty of wrong ways to do a chatbot, enter Ryanair “A passenger named Tom tweeted, “So @Ryanair decided to give me a voucher despite being in the queue for a refund for a month! And if I want to get a proper refund, I have to wait in line to speak to a chatbot, who is only available during the day. Give me back my money. ”
The Future of Bots
Even as current generation bots are rolled out to support today’s needs, the bots themselves are getting smarter and more flexible to deliver more practical, flexible solutions in the coming years.
Take Facebook’s new Blender chatbot which has thrown down the gauntlet to Google’s Meena. The two are seemingly embroiled in their own spec/feature war that would put AMD vs Intel or Sony vs Nintendo to shame, its good to see a largely pointless bragging war in tech but it also indicates how seriously the players are taking the future of bots and AI.
Adding fuel to the fire, Facebook researchers recently demonstrated Situated Interactive MultiModal Conversations (SIMMC), helping train AI chatbots in specific markets. In a research paper (PDF), they detail new data sets designed to deliver next-generation virtual assistants that can handle multimodal inputs (e.g., smartphone camera vision, memories of previous interactions, in addition to the user’s inputs and speech), and perform multimodal actions. The tests used around 13,000 human-to-human dialogs across two domains — furniture and fashion — along with several tasks framed as objective evaluation protocols to deliver appropriate answers.