Around the world, sports fans are being kept away from stadia, tracks and courses – socially distanced from their teams and idols. Chatbots were playing a key role in helping keep them in touch and engaged, but not all brands are doing enough and there’s a big gap between where sports bots need to be, with AGI likely to help get them over the line.
Since the rise of the common chatbot, sports brands have been trying them out in various ways to draw in the massive audience. Progress has been picking up, and last November the NBA basketball league launched an AI chatbot, #NBAChat as part of the drive to engage more on social media. That was before COVID and the simple aim of using Facebook Messenger to deliver in-game highlights seemed like a natural progression from team bots selling tickets and merchandise.
The NBAChat chatbot provided news updates, previews, notifications, score alerts, Then, along came COVID, crippling the new season and leaving fans in hoops limbo. The previous season, Australia’s AFL league launched a bot to interact with fans, seeing some 50,000 fans chat, with some 15,000 active users per week, helping to attract younger fans especially. And then the virus hit, and their 2020 season was suspended in March.
What did both of these massive sporting brands do with their bots during the crisis? Nothing! There’s no discernable change in messaging or function, even as both sports on different sides of the world had plenty of urgent messages to get out to supporters. At the same time, the rest of the business, government and health world were adopting bots at a huge scale.
They shared urgent advice, promoted best practics and got updates out as information and schedules changed fast. Where were expensive marketers for giant sports teams and leagues? They certainly were not getting furloughed!
At least the NBA has added a link to its Restart page with the whole truncated 88 game season taking place at the ESPN Sports Complex in Orlando, Florida. But, it still misses the engagement opportunity that millions of fans will be looking for, with their questions and need for information.
Making a better stab of things is the NHL with their Return to Play resumption of the league a part of the NHL’s chatbot, but these bots with such powerful reaches, really aren’t doing the sports any favours or really trying to engage with fans at a time when those fans need it the most.
The future for sports bots
Perhaps these brands are waiting for the artificial general intelligence bots (AGI) from the likes of Facebook and Google? AGI bots will be able to discuss pretty much any subject and provide human-like conversation.
That would save bot developers having to put in lots of extra training and work for their current generation of bots. And they would be able to promote something that would be much more impressive for future seasons, when there’s more of a normal feel to events.
The road to AGI promises bots that can deliver on the sci-fi promise of AI that both understands what you ask it, on any subject, and can provide an answer in context to the conversation. Facebook and Google among others are working on these AGI bots, and while many businesses won’t need them, their flexibility could be the things sporting big-names are after to make for more engaging conversations with fans.
Facebook claims that its tool. Blender, is “the largest-ever open-domain chatbot, outperforming others in terms of engagement and also feels more human.” While Google’s Meena is being developed into a “conversational agent that can chat about…anything.”
Both are stretching what we currently recognise as AI, with expansive training. Facebook’s bot uses a model with 9.4 billion parameters, but both the giants and many other bot builders will be aiming to deliver creations that understand their audience, so while your typical sports fan might bot care about the AI behind a bot, they will appreciate a good chat about new signings, prospects in their next match, as well as the usual player chat and statistics.
Google aims to meet that need by measuring Perplexity, “Perplexity measures the uncertainty of a language model. The lower the perplexity, the more confident the model is in generating the next token (character, subword, or word). Conceptually, perplexity represents the number of choices the model is trying to choose from when producing the next token.”
That and better natural language understanding (NLU) and the ability to translate text to speech so bots can better talk via smartphones, smart home gadgets like Amazon Echo and so on, will make the sports fan more accepting of bots in their environment.
The key to success will be what the bot offers to fans. Since we’re a little way from building or using AGI bots, here we have to play with some scenarios. For a start, from the human perspective, there are so many nicknames for teams and players that bots will need to understand the context and use of them.
- Bots might recognise a fan using older nicknames that indicates their interest in a different era of players and look to compare them to modern squad members.
- Bots will need to understand the conversation behind statistics, rather than just regurgitate roster numbers. With AI and computational power behind it, an AGI bot could compare the likely performance of two or more players for the game or season ahead and discuss strengths and weaknesses.
- Some fans will love to trash talk, even with bots. And the bot will need to be ready to respond in a way that doesn’t offend but enters the spirit of sporting “debate.” AGIs will understand what qualifies as banter and learn to ignore deliberate provocation while going along with fun chats.
Those are just three ways AGI bots can help deliver a better conversation with sports fans around the world, when the doors open. From soccer stadiums to ice hockey areas, the next world cups or big athletics events, bots need to be bigger and better, as well as ready to respond to an audience that is now more used to engaging with bots.
The smarter bots will also be better equipped to deal with ticket bookings, event date changes, and when the next crisis comes along will also be able to back up the good advice coming from other agencies to help spread the word among fans. So, while there’s plenty of time for these potentially massive bots to go through development, the likes of F1, the NFL NASCAR, need to get a move on. Japan’s bots for the 2020 Olympics might be delayed to 2021, but imagine how much better their bots will be with a little more time.
As with business, not all sporting events or teams will need a fully-powered AGI bot, but for the big leagues and brands, once one does the others will need to be in step to talk to their digital audience.