- Tech has come far but still some ways to go to be truly intelligent
- Still, there are pockets of opportunities for enterprises to leverage
I DON’T normally keep that close an eye on the announcements per se from The Consumer Electronic Show (CES) held in Las Vegas recently because of its sheer volume and variety but this year I did, as artificial intelligence (AI) somewhat dominated many of the headlines that came out of the show.
On the first day of the show, Lenovo introduced a clone-like device to the hugely successful and popular virtual assistant Amazon Echo. Dubbed Lenovo Smart Assistant, it is able to take commands much like the Echo does.
Then graphics giant Nvidia announced an AI-based system called Nvidia co-pilot. The aim is to diversify itself from its graphics roots and partner with car manufacturers like Audi to develop AI for cars.
Next up was what I considered to be the most interesting development from CES. China’s Huawei Technologies Co Ltd surprised everyone by announcing that its latest Mate 9 smartphone will come pre-installed with an app that gives users access to Amazon.com’s virtual assistant called Alexa, found in its Echo devices, albeit only for the US market.
The Verge reports that Huawei claims the app will offer customers a “natural, convenient user interface” for talking to the digital assistant, and will be capable of all the usual tasks, including setting alarms, making to-do lists, and getting information about the news and weather.
Leaving aside the details of the implementation and why Huawei only targeted the US market, this development is interesting in that we’re beginning to see a greater proliferation of virtual assistants making their way to perhaps the most ubiquitous device on the planet – the smartphone.
You may argue that virtual assistants are not new, and that’s perfectly true. Apple’s Siri has been on the market for years and so have Google’s Google Assistant and Microsoft’s Cortana.
It’s not my intent here to compare how good these services are or which one outperforms the other. For that, you can read the many websites, which have done comparisons between the various services.
For me though, I’m more interested to explore what these virtual assistant can mean to the larger business world and the impact they bring to areas such as customer service and sales enablement, especially now that they are appearing at the consumer level via smartphones.
For the record, Amazon’s Alexa wasn’t just found in Huawei’s Mate 9 but it was a huge focus having made its way into a slew of devices made by a number of consumer electronic giants, a full list of which can be found here.
A primer on chatbots
Virtual assistants can also be termed as chatbots in industry parlance. But what really are chatbots? In the simplest definition and for the purpose of this column, chatbots are software that have been coded to simulate human conversations or interactions that happen on devices and robots.
Chatbots typically can simulate basic conversations through automated and structured responses that is triggered by an input, say, when someone says something. They can also be interfaces (physical or virtual) that provide intelligent interactions and continue to improve based on more experience and better data.
According to natural language processing researcher Dr Alyona Medelyan (pic, above), chatbots have come a long way since the early days, as software and algorithms today are very much more advanced.
Noting that chatbots today are more of a platform that can be programmed to learn and where developers can add capabilities, chatbots can gradually improve their skills, explained the CEO of her own startup Thematic.
“Chatbots are effective today at figuring out user intent in a specific context more than they did before, and so in that sense, it they have improved tremendously from the past,” she told Digital News Asia (DNA) in an interview.
However as far as chatbots have come, Medelyan conceded that it’s still early days for the technology.
Chatbots are powered by a variety of technologies. They incorporate deep machine learning as well as linguistic skills such as natural language processing, Medelyan explained.
In deep machine learning, which is a subset of AI, machines are taught to recognise things like patterns, phrases, and images repeatedly in a bid to accustom them to eventually provide the right answers to a given set of questions, she noted.
“But deep learning is not deep understanding,” she argued. “Deep learning is just optimisation of software algorithms, which are smart enough to make connections given enough data.”
According to Medelyan, language today is still extremely difficult for a computer to figure out today as much of it depends on context and nuances.
“The main difficulties is around semantics, the understanding of the meaning of words,” she explained. “Saying a phrase like ‘I thirst” could have different meaning depending on the context. Also same words can have different meanings in different contexts. This presents difficulty for machines as they constantly need to figure out the nuances and context.”
Medelyan said many of today’s chatbots are programmed to listen to keywords in a specific context. For example in a customer service situation, a chatbot is programmed to listen to words such as ‘price,’ ‘opening hours/times,’ ‘address,’ ‘contact number’ et al.
“Chatbots today are optimised for specific use cases and they can be effective that way,” she argued. “But there is still some way to go to have a greater understanding of semantics and context.”
Chatbots in the enterprise
Still, the prospect of using chatbots in specific sectors of the industry has great potential despite the limitations for now. One of the leaders in this field, Amazon Web Services (AWS), recently announced a service known as Amazon Lex, which the company believes has the potential to take chatbots to the next level.
According to AWS product marketing manager Lowell Anderson (pic, above), Lex allows anyone to build conversational interface into any types of apps.
Speaking to Digital News Asia at its recently concluded annual re:Invent 2016 conference. Anderson said that with the introduction of Lex, it becomes easy for developers to build applications such as a mobile sales force app that has a conversational interface to pull data from their database when they’re out in the field.
“A sales force guy can query what is the last thing the person bought or when is their contract up, just by voice instead of having to type these into a tablet,” he claims. “The same can be done with an inventory management person wanting to know what inventory levels are like in the field by voice, which is much more powerful and convenient.”
Anderson said AWS is currently building ‘connectors,’ pieces of software that can connect Lex to common services such as customer relationship management (CRM) software Salesforce.com.
“We are delivering a service that makes it easy for iOS and Android developers to build mobile backend, using the AWS mobile hub service, so it’s straightforward to put together the mobile backend that leverages Lex as well as services like Amazon S3 and mobile analytics.”
AWS’ services is only but an example of the things to come for chatbots in the enterprise.
In fact, according to studies conducted by Tech Research Asia chatbots are already here today and are being used in various capacities. For instance, Chinese search giant, Baidu, launched a health chatbot called Melody for patients and doctors.
Singapore-based bank DBS, released a conversation agent, which customers can use on their favorite mobile messaging app. Just Google other examples, you’ll find more.
There are other potential use cases for chatbots appearing but many of them are tied to interaction with customers. For example, a wealth or financial chatbot advisory service set up by a bank that advises customers in multiple languages in multiple countries.
Or a complaints or claims handling chatbot, which is able to interface into a company’s social media channels and is constantly ‘listening’ to what customers and would-be customers are saying about a product or service. They can then recommend an appropriate response to those observations.
Finally, a marketing or sales generation lead chatbot set up by a company, which not only welcomes customers to the site but is able to pleasantly capture contact details, and provide access to the content on a voice-assisted basis.
Salient points to note
While I concede that things are still fluid and nascent, the aforementioned points do demonstrate that virtual assistants or chatbots are the next frontier for the industry.
As such, the first thing organisations need to do is to seriously take note of these developments and not merely dismiss them as being too futuristic for them to deploy as such notions will be to their detriment in the long run.
Secondly, the use of chatbots aren’t tied to a specific geography or industry sector as its deployment is still very new and there aren’t many use cases as yet. Typically, business in South-East Asia (SEA) lag deployment and use case when compared with western countries.
But with chatbots, countries in SEA aren’t necessarily hemmed in by this paradigm. The most innovative companies even here in SEA can therefore lead the way and set trends for others to follow. With chatbots, you potentially have the ability to build new business models and outcomes from them.
That said, companies shouldn’t rush headlong into deploying chatbots just for the ‘coolness’ factor. Deploying chatbots can potentially change the way you interact with customers both positively and also negatively, if not done correctly. It could enhance your brand as a leader or disrupt your brand as a company who doesn’t know what it’s doing.
There are also issues to do with new set of technologies, services and processes that need to be managed effectively and if not done properly can affect your very core culture for doing business.
Those planning for chatbots should think about providing an engaging and accurate two-way interaction in an automated way as much as possible. At the same time, no system is completely automated and must be backed up by human support for the best economics per customer transaction or interaction.
The use of chatbots also isn’t ‘mono-directional’ in that every transaction can also provide feedback for a company to refine their processes in order for them to engineer more accurate information or advice to customers (or employees), while also contributing insights on those same individuals that the organisation can use.
This could be perhaps the most valuable derivation from chatbots because it has the potential to capture detailed information about people, their interactions, their preferences such that companies can refine their targeted offering to customers.
Finally, a chatbot deployment isn’t a standalone piece and must be seamlessly integrated into an omni-channel management that continues to allow the organisation to have advanced reporting, auditing and incident response management.
Edwin Yapp is contributing editor to Digital News Asia and Asean analyst at Tech Research Asia, an advisory firm that translates technology into business outcomes for executives in Asia Pacific.
The inevitable rise of the robots
IoT predictions for 2017: What’s the buzz?
Amazon Web Services trains its spotlight on artificial intelligence
For more technology news and the latest updates, follow us on Twitter, LinkedIn or Like us on Facebook.