Our November 2022 Event: A Glimpse into the World of Natural Language Processing. A Talk by Anna Koroleva

Artificial Intelligence (AI) is one of the more controversial topics in linguistics, with some saying that AI is a useful tool and others worried that linguists will be replaced by AI in the future. On November 25th, we were joined by Computational Linguist, Anna Koroleva, to discuss natural language processing (NLP) and why she believes it is not coming to replace us.

Anna started off the talk by explaining how NLP is a sub-category of AI, and defined NLP as:

A field of study that aims to process a large amount of text (unstructured data) to extract structured data that can be further analysed to gain valuable insights on a given topic.

However, perhaps the easiest way to understand NLP is the examples of its use in daily life. For example, search engines such as Google, and machine translations such as Google Translate. Anna explained how these tools have been developed through NLP, such as Google collecting all the relevant information on a topic that we have typed into the search bar. NLP allows us to communicate with AI by developing these programmes to analyse and understand natural language data.

Anna also discussed other examples of NLP, such as spell check and voice assistants that we may use on an everyday basis.

Other Uses of NLP

Anna then explained how NLP can be used in healthcare, for example to answer medical questions. Doctors are able to search for a specific aspect of healthcare and instantly retrieve all the relevant information, instead of having to retrieve the answers manually, and search through thousands of articles for relevant information. Another example was the use of chatbots to answer questions about mental health – which was then discussed further in the Q&A section.

Another field that can benefit from NLP is forensic linguistics. Anna explains how NLP can be used to analyse two documents, and determine if they were written by the same author based on writing style. This is a task that is often done manually, but can be automated using NLP to save time.

Finally, NLP can also be used for marketing, to compare different marketing campaigns and analyse which writing practices have been used in more successful campaigns – helping to dictate the best practices for future campaigns. For example, if a successful campaign uses a lot of 2nd person pronouns, then you will know to keep using them in future campaigns.

Will NLP/AI Replace Human Professionals?

We then moved on to the topic of potential negative implications for NLP, and how some people believe it will replace human linguists. Anna started with an example of a machine error, explaining how her surname ‘Koroleva’ can also mean ‘queen’ in Russian when spelled in the lowercase ‘королева’ – meaning that Google translates her name as ‘Queen Anne’. This demonstrates how NLP is still imperfect compared to human translators.

Anna then explained how the areas where AI is useful does not necessarily overlap with human professional work. For example, the situations in which we’d use Google Translate are not the same situations that we need a translator for. Google Translate can be useful when travelling, but professional translation work such as legal translation is too important to risk using AI that could make mistakes that a human wouldn’t.

There are also many professions in fields such as healthcare and forensic linguistics where NLP can assist the work done by human professionals, but the decision-making needs to be done by humans. This is once again because the implications of a machine error are too important, as this could lead to incorrect diagnoses or wrong convictions.

Q&A

After the talk, we took questions from the attendees, which led to many fascinating discussions taking place.

The first question was about how a Computational Linguist introduces themselves and explains their profession to people. Anna agreed that the phrase ‘Computational Linguist’ is often confusing to people, and stated that it’s simply easier to describe her work as ‘text analytics’ as this can often be easier for people to understand.

The next question was open to everyone, asking what we would like to ask a chatbot that cannot currently be answered. This received many creative responses, such as ‘will AI take over the world?’ ‘What is the meaning of life?’ And ‘why are some people so mean?’ This linked back to the earlier topic of chatbot’s helping with mental health, and the person who asked the question revealed that they are working on a mental health chatbot at the moment.

Other topics discussed in the Q&A included switching careers to the NLP field, black box deep learning models and the struggles of NLP with small datasets, just to name a few.

Conclusion

The aims of this talk were to introduce NLP to those unfamiliar, provide information on its practical uses for language professionals and discuss further issues surrounding NLP and AI in general. We would like to thank everyone who joined and participated, and we hope to have succeeded in all three areas.

Special thank you as well to Anna for the great talk, and for reassuring us that our jobs may be safe from AI.

Language professionals’ work can be assisted and made easier by NLP, but they can’t be fully replaced.

About the Speaker

Anna Koroleva

Anna is a computational linguist/NLP Data Scientist. She holds an MSc in Theoretical and Applied Linguistics and a PhD in Natural Language Processing. She has 9 years of research and industrial experience in NLP. Her professional interests include Biomedical NLP, Conversational AI and AI for Good.

Feel free to reach out to Anna via LinkedIn.

---

Any questions or comments on NLP? Leave a comment below and let’s continue the conversation there.