Natural language processing is still being refined, but its popularity continues to rise. This new, better version is likely to help.
When you speak to a computer, whether on the phone, in a chat box, or in your living room, and it understands you, that’s because of natural language processing. The computer voice can listen and respond accurately (most of the time), thanks to artificial intelligence (AI).
SEE: Hiring kit: Data Scientist (TechRepublic Premium)
Natural language processing (NLP) is the language used in AI voice questions and responses. The processing of language has improved multi-fold over the past few years, although there are still issues in creating and linking different elements of vocabulary and in understanding semantic and contextual relationships.
Despite these continued efforts to improve NLP, companies are actively using it. NLP has been a hit in automated call software and in human-staffed call centers because it can deliver both process automation and contextual assistance such as human sentiment analysis when a call center agent is working with a customer.
NLP has also been used in HR employee recruitment to identify keywords in applications that trigger a close match between a job application or resume and the requirements of an open position.
SEE: An IT pro’s guide to robotic process automation (free PDF) (TechRepublic)
In our homes, we use NLP when we give a verbal command to Alexa to play some jazz. So there’s no surprise that NLP is on nearly every organization’s IT road map as a technology that has the potential to add business value to a broad array of applications.
This is precisely why the recent breakthrough of a new AI natural language model known as GPT-3. is significant.
What is GPT-3?
With GPT-3, 175 billion parameters of language can now be processed, compared with predecessor GPT-2, which processes 1.5 billion parameters. This new GPT-3 natural language model was first announced in June by OpenAI, an AI development and deployment company, although the model has not yet been released for general use due to “concerns about malicious applications of the technology.”
SEE: IBM highlights new approach to infuse knowledge into NLP models (TechRepublic)
“GPT-3 takes the natural language Transformer architecture to a new level,” said Suraj Amonkar, fellow AI@scale at Fractal Analytics, an AI solutions provider. “It’s built for all of the world’s languages, and has machine translation.”
The possibilities with GPT-3 are enticing.
- For a government or a multinational corporation, the ability to rapidly localize text and voice-based messages or translate them into virtually any world language—and to do it with automation—opens access to new customers and better support for field offices in foreign countries that are supporting company products or operations.
- For research institutions and for medical and life sciences researchers, the ability to easily translate a paper that is written in a foreign language can be done rapidly.
- For media, publishing, and entertainment companies, there can be a fast way to translate the spoken and written word into many different languages.
How GPT-3 can help an organization
If you’re looking at the IT strategic road map, the likelihood of using or being granted permission to use GPT-3 is well into the future unless you are a very large company or a government that has been cleared to use it, but you should still have GPT-3 on your IT road map.
There is also a strong argument that if you are the CIO of a smaller organization, that the evolution of NLP language modeling into GPT-3 capabilities should not be ignored because natural language processing and the exponential processing capabilities that GPT-3 language modeling endows AI with are going to transform what we can do with processing and automating language translations and analytics that operate on the written and spoken word.
If you’re doing business in a global economy, as almost everyone is, that capability will be invaluable.