How AI Has Changed Natural Language Processing

In our digital-driven, hyperconnected world, businesses using data-driven approaches to make their decisions is the new norm. And while there's plenty of data through which businesses can parse to arrive at operational decisions, it's not always clear how to best use the value that data contains.

Artificial intelligence (AI) has, for a long time, been considered the best processor to tease out that value. When it comes to creative and cognitive-driven tasks, though, it's often fallen short - that is, until recently. The huge strides speech-based AI has made in the past few years have challenged the typical notion that AI is a limited technology. Nowhere is this more apparent than in the AI subfield called "natural language processing" (NLP).

Natural language processing is a computer's ability to make sense of the meaning behind speech or text, just like a human being. Ever asked Siri or Alexa to answer a question for you? If so, you've already had your first taste of NLP in action. However, NLP has made its way into the business world, too.

Let's quickly cover what NLP can do for businesses, how AI has transformed NLP, and how that may affect businesses in the future.

How NLP can help businesses

When it comes to NLP tools, the one that's most widely known is AI research lab OpenAI's GPT-3. Assignable to pretty much any language-based task, GPT-3 can fill in blanks by predicting words and even finish written narratives based on just one sentence sample.

Gurus of NLP refer to processing tools like GPT-3 as language models.

Language models can accomplish simple tasks like categorizing documents and sentiments from paragraphs, but they're also powerful enough to respond to queries and paraphrase written reports. Imagine that you started a new business that just launched its website, and you want to summarize to your visitors how you're protecting your website infrastructure and keeping their data safe.

With the power of language models, you could distill the ideas of your infrastructure engineers into summaries that even a non-specialist could understand.

While you're building that website of yours, you realize you'll need to hire a backend developer or two to put together the foundation and core functions behind your site, so you get to work on putting together a job posting.

Rather than spend lots of time doing research yourself to discover that the average hourly wage for an intermediate backend developer is around $60, you could use a language model to parse through salary statistics pages and summarize the information you need. And, thanks to AI, different large language models and the responses they generate are becoming more closely aligned with the values and intentions that we as human beings hold.

Benefits to jobs, programming, and even the economy

AI research and deployment have helped make language models like GPT-3 particularly promising when it comes to coding and writing - you only need to look to OpenAI's GPT-3-based model Codex for a prime example. Codex helps programmers do their jobs more efficiently by ingesting natural language input and outputting code. It can also craft rudimentary video games and power products such as Microsoft's code suggestion service, Copilot.

Anticipation has already been high that language models that generate code will make programmers' lives easier. AI research and deployment have surpassed even the biggest expectations riding on this kind of transformative capability. British AI research lab DeepMind, for example, released a language model demonstrating cognitive reasoning and decision-making that can leave humans in the dust during programming competitions.

AI researchers and NLP experts consider language models such as GPT-3 "foundation models."

This is only a recent area of research in AI, and it pertains to a model's ability to work not just with text but with video and images as well. Certain models can even be trained in different data formats concurrently, and foundation models' rapid transformation of the cognitive processing they perform has some economists positing that it may influence economic growth on par with the industrial revolution.

How businesses can prepare for AI in the future

You're likely keenly aware of just how valuable data is, but what you may not know is that you're probably glancing over some key data sets if you're not using NLP across your business. Just picture the myriad text data assets you interact with daily, like emails, reports, press releases, and phone calls and meetings. With the help of text analytics and NLP, you can transcribe these text assets and save yourself - and your colleagues - lots of valuable time.

You'll also want to understand how to best use AI-based tools to drive more informed decisions and reorganize your labor. Although strides in AI won't get rid of jobs outright, language-based AI is poised to automate many tasks (including decision-making tasks). AI-based tools that allow people across your organization to arrive at data-informed decisions can help you change the nature of your programmers' and developers' roles.

If we look back to Codex for a moment, we can anticipate that the future of reorganized work may include a smaller number of dedicated programmers and a greater number of employees who can put modest programming skills to use.

Conclusion

Artificial intelligence has paved the way for businesses to leverage language-based tools in multiple ways. Multimodal tools based on foundation models have the potential to reshape the way organizations do business in ways that are still uncertain.

Though the future of language-based AI tools may still be nebulous, what's certain is that you can begin identifying your organization's text assets and start understanding the different cognitive tasks that your organization's roles tackle every day. It's essential that you begin to adopt language-based tools, even if some may prove more useful than others. The quicker you do this, the clearer the nature of your organization's future roles will become.