What can NLG do for your business?
Natural Language Generation (NLG) is the lesser known artificial intelligence (AI) technology that translates raw data into understandable text or spoken word. It’s also one of the most interesting of AI’s upcoming applications for business.
In an exclusive interview with TechHQ, Apoorv Saxena, Product Manager for Cloud AI & Google AI discusses how businesses can benefit from NLG:
How is NLG technology most commonly used?
The Google Assistant is an example of NLG in action; for example, how it answers questions about yesterday’s winning soccer team or summarizes the text in an email.
There is also a lot of research being conducted on NLG to generate news articles, especially in the financial services space.
To date, however, the technology has been most successful with structured data, e.g. summarizing an earnings report based on revenue reported.
From that type of data, NLG can generate a very specific template which says company X released its earnings and surpassed analysts’ predictions by X%.
Another extensive application of NLG is in chatbots.
Essentially, they can be seen as a version of NLG where people, and in this case [Google] customers, are programming the bot to respond to a user’s query in a very specific way.
What challenges does development of the technology present?
NLG is extremely complex, creating a comprehensible narrative around multiple facts is challenging.
The technology has seen great advancements in the last few years but there is still much work to be done, especially around customization and conversation; that is where a lot of the progress will be made over time.
For example, for it to work for customized responses, specific templates need to be created based on new use cases as the complexity increases.
Improving the various tones, emotions, and expressions used in NLG is the next step in research.
A simple question like: “What is the temperature today?” could generate countless versions of the same answer, such as ‘hey the weather is 25 degrees’ and ‘it’s going to get hotter’ or ‘it’s going to be 25 degrees today, no need for an umbrella’.
Enabling conversation is another development challenge for NLG as it adds a layer of complexity which is reasoning and that is essentially not seen as part of NLG.
What are the future applications of NLG?
Like any other AI technology, we predict NLG will improve the productivity and scale of what it can do. Imagine a small business is dealing with a lot of customer queries and they can’t answer them all – with NLG a chatbot can answer these queries.
This is what we see as augmenting existing human input or improving productivity and allowing businesses to scale significantly.
In some cases, if a customer must wait 15 minutes to speak with an agent they would rather talk to a bot, but if customers can get to a human agent they always prefer to talk to a human. It’s still too early to say how NLG will be used as it is still relatively new.
Compared to other AI technologies, where does this stand in significance and value?
The value of NLG depends on the users being targeted, for example, robotic process automation (RPA) is being widely used in the financial services sector to automate customers onboarding and is very useful in that case but not others.
These powerful technologies are going through a significant upgrade cycle in terms of capabilities and I think customers are still discovering new use cases.
Do businesses need to invest in different AI technologies available or is it better to use them through one simple platform?
Large enterprises who have the capital and the talent will invest in NLG and integrate it into their core products and business processes.
Medium and small businesses will consume NLG, not in its raw form, but will integrate it into another platform.
For instance, almost every SaaS company is investing heavily in integrating AI into their products.
They are doing this to boost their core business processes because it helps them be more productive and improve the customer experience. Our view is that NLG is evolving quickly and there will be a lot more opportunities in the future.
- A year of high-severity attacks and groundbreaking cybersecurity strategies in 2023
- How default probability analytics can make a difference
- SMIC defying US sanctions with 5nm innovation and Huawei alliance
- Nvidia’s CEO, Jensen Huang: AI will take over coding, making learning optional
- Chinese cloud companies in pricing war as Alibaba slashes prices