Will AI Replace Jobs? 17 Job Types That Might be Affected

What Is Natural Language Processing? - EWeek

eWEEK content and product recommendations are editorially independent. We may make money when you click on links to our partners. Learn More.

Natural language processing (NLP) is a branch of artificial intelligence (AI) that focuses on computers incorporating speech and text in a manner similar to humans understanding. This area of computer science relies on computational linguistics—typically based on statistical and mathematical methods—that model human language use.

NLP plays an increasingly prominent role in computing—and in the everyday lives of humans. Smart assistants such as Apple's Siri, Amazon's Alexa and Microsoft's Cortana are examples of systems that use NLP.

In addition, various other tools rely on natural language processing. Among them: navigation systems in automobiles; speech-to-text transcription systems such as Otter and Rev; chatbots; and voice recognition systems used for customer support. In fact, NLP appears in a rapidly expanding universe of applications, tools, systems and technologies.

In every instance, the goal is to simplify the interface between humans and machines. In many cases, the ability to speak to a system or have it recognize written input is the simplest and most straightforward way to accomplish a task.

While computers cannot "understand" language the same way humans do, natural language technologies are increasingly adept at recognizing the context and meaning of phrases and words and transforming them into appropriate responses—and actions.

Also see: Top Natural Language Processing Companies

Natural Language Processing: A Brief History

The idea of machines understanding human speech extends back to early science fiction novels. However, the field of natural language processing began to take shape in the 1950s, after computing pioneer Alan Turing published an article titled "Computing Machinery and Intelligence." It introduced the Turing Test, which provided a basic way to gauge a computer's natural language abilities.

During the ensuing decade, researchers experimented with computers translating novels and other documents across spoken languages, though the process was extremely slow and prone to errors. In the 1960s, MIT professor Joseph Weizenbaum developed ELIZA, which mimicked human speech patterns remarkably well. Over the next quarter century, the field continued to evolve. As computing systems became more powerful in the 1990s, researchers began to achieve notable advances using statistical modeling methods.

Dictation and language translation software began to mature in the 1990s. However, early systems required training, they were slow, cumbersome to use and prone to errors. It wasn't until the introduction of supervised and unsupervised machine learning in the early 2000s, and then the introduction of neural nets around 2010, that the field began to advance in a significant way.

With these developments, deep learning systems were able to digest massive volumes of text and other data and process it using far more advanced language modeling methods. The resulting algorithms had become far more accurate and utilitarian.

Also see: Top AI Software 

How Does Natural Language Processing Work?

Early NLP systems relied on hard coded rules, dictionary lookups and statistical methods to do their work. They frequently supported basic decision-tree models. Eventually, machine learning automated tasks while improving results.

Today's natural language processing frameworks use far more advanced—and precise—language modeling techniques. Most of these methods rely on convolutional neural networks (CNNs) to study language patterns and develop probability-based outcomes.

For example, a method called word vectors applies complex mathematical models to weight and relate words, phrases and constructs. Another method called Recognizing Textual Entailment (RTE), classifies relationships of words and sentences through the lens of entailment, contradiction, or neutrality. For instance, the premise "a dog has paws" entails that "dogs have legs" but contradicts "dogs have wings" while remaining neutral to "all dogs are happy."

A key part of NLP is word embedding. It refers to establishing numerical weightings for words in specific context. The process is necessary because many words and phrases can mean different things in different meanings or contexts (go to a club, belong to a club or swing a club). Words can also be pronounced the same way but mean different things (through, threw or witch, which). There's also a need to understand idiomatic phrases that do not make sense literally, such as "You are the apple of my eye" or "it doesn't cut the mustard."

Today's models are trained on enormous volumes of language data—in some cases several hundred gigabytes of books, magazines articles, websites, technical manuals, emails, song lyrics, stage plays, scripts and publicly available sources such as Wikipedia. As the deep learning system parse through millions or even billions of combinations—relying on hundreds of thousands of CPU or GPU cores—they analyze patterns, connect the dots and learn semantic properties of words and phrases.

It's also often necessary to refine natural language processing systems for specific tasks, such as a chatbot or a smart speaker. But even after this takes place, a natural language processing system may not always work as billed. Even the best NLPs make errors. They can encounter problems when people misspell or mispronounce words and they sometimes misunderstand intent and translate phrases incorrectly. In some cases, these errors can be glaring—or even catastrophic.

Today, prominent natural language models are available under licensing models. These include the OpenAI codex, LaMDA by Google, IBM Watson and software development tools such as CodeWhisperer and CoPilot. In addition, some organizations build their own proprietary models.

How is Natural Language Processing Used?

There are a growing array of uses for natural language processing. These include:

Conversational AI. The ability of computers to recognize words introduces a variety of applications and tools. Personal assistants like Siri, Alexa and Microsoft Cortana are prominent examples of conversational AI. They allow humans to make a call from a mobile phone while driving or switch lights on or off in a smart home. Increasingly, these systems understand intent and act accordingly. For example, chatbots can respond to human voice or text input with responses that seem as if they came from another person. What's more, these systems use machine learning to constantly improve.

Machine translation. There's a growing use of NLP for machine translation tasks. These include language translations that replace words in one language for another (English to Spanish or French to Japanese, for example). Google Translate and DeepL are examples of this technology. But machine translation can also take other forms. For example, NLP can convert spoken words—either in the form of a recording or live dictation—into subtitles on a TV show or a transcript from a Zoom or Microsoft Teams meeting. Yet while these systems are increasingly accurate and valuable, they continue to generate some errors.

Sentiment analysis. NLP has the ability to parse through unstructured data—social media analysis is a prime example—extract common word and phrasing patterns and transform this data into a guidepost for how social media and online conversations are trending. This capability is also valuable for understanding product reviews, the effectiveness of advertising campaigns, how people are reacting to news and other events, and various other purposes. Sentiment analysis finds things that might otherwise evade human detection.

Content analysis. Another use case for NLP is making sense of complex systems. For example, the technology can digest huge volumes of text data and research databases and create summaries or abstracts that relate to the most pertinent and salient content. Similarly, content analysis can be used for cybersecurity, including spam detection. These systems can reduce or eliminate the need for manual human involvement.

Text and image generation. A rapidly emerging part of natural language processing focuses on text, image and even music generation. Already, some news organizations produce short articles using natural language processing. Meanwhile, OpenAI has developed a tool that generates text and computer code through a natural language interface. Another OpenAI tool, dubbed Dall-E-2, creates high quality images through an NLP interface. Type the words "black cat under a stairway" and an image appears. GitHub Copilot and Amazon CodeWhisperer can auto-complete and auto-generate computer code through natural language.

Also see: Top Data Visualization Tools 

NLP Business Use Cases

The use of NLP is increasingly common in the business world. Among the top use cases:

Chatbots and voice interaction systems. Retailers, health care providers and others increasingly rely on chatbots to interact with customers, answer basic questions and route customers to other online resources. These systems can also connect a customer to a live agent, when necessary. Voice systems allow customers to verbally say what they need rather than push buttons on the phone.

Transcription. As organizations shift to virtual meetings on Zoom and Microsoft Teams, there's often a need for a transcript of the conversation. Services such as Otter and Rev deliver highly accurate transcripts—and they're often able to understand foreign accents better than humans. In addition, journalists, attorneys, medical professionals and others require transcripts of audio recordings. NLP can deliver results from dictation and recordings within seconds or minutes.

International translation. NLP has revolutionized interactions between businesses in different countries. While the need for translators hasn't disappeared, it's now easy to convert documents from one language to another. This has simplified interactions and business processes for global companies while simplifying global trade.

Scoring systems. Natural language is used by financial institutions, insurance companies and others to extract elements and analyze documents, data, claims and other text-based resources. The same technology can also aid in fraud detection, financial auditing, resume evaluations and spam detection. In fact, the latter represents a type of supervised machine learning that connects to NLP.

Market intelligence and sentiment analysis. Marketers and others increasingly rely on NLP to deliver market intelligence and sentiment trends. Semantic engines scrape content from blogs, news sites, social media sources and other sites in order to detect trends, attitudes and actual behaviors. Similarly, NLP can help organizations understand website behavior, such as search terms that identify common problems and how people use an e-commerce site. This data can lead to design and usability changes.

Software development. A growing trend is the use of natural language for software coding. Low-code and no-code environments can transform spoken and written requests into actual lines of software code. Systems such as Amazon's CodeWhisperer and GitHub's CoPilot include predictive capabilities that autofill code in much the same way that Google Mail predicts what a person will type next. They also can pull information from an integrated development environment (IDE) and produce several lines of code at a time.

Text and image generation. The OpenAI codex can generate entire documents, based a basic request. This makes it possible to generate poems, articles and other text. Open AI's DALL-E 2 generates photorealistic images and art through natural language input. This can aid designers, artists and others.

Also see: Best Data Analytics Tools 

What Ethical Concerns Exist for NLP?

Concerns about natural language processing are heavily centered on the accuracy of models and ensuring that bias doesn't occur. Many of these deep learning algorithms are so-called "black boxes," meaning that there's no way to understand how the underlying model works and whether it is free of biases that could affect critical decisions about lending, healthcare and more.

There is also debate about whether these systems are "sentient." The question of whether AI can actually think and feel like a human has been expressed in films such as 2001: A Space Odyssey and Star Wars. It also reappeared in 2022, when former Google data scientist Blake Lemoine published human-to-machine discussions with LaMDA. Lemoine claimed that the system had gained sentience. However, numerous linguistics experts and computer scientists countered that a silicon-based system cannot think and feel the way humans do. It merely parrots language in a highly convincing way.

In fact, researchers who have experimented with NLP systems have been able to generate egregious and obvious errors by inputting certain words and phrases. Getting to 100% accuracy in NLP is nearly impossible because of the nearly infinite number of word and conceptual combinations in any given language.

Another issue is ownership of content—especially when copyrighted material is fed into the deep learning model. Because many of these systems are built from publicly available sources scraped from the Internet, questions can arise about who actually owns the model or material, or whether contributors should be compensated. This has so far resulted in a handful of lawsuits along with broader ethical questions about how models should be developed and trained.

Also see: AI vs. ML: Artificial Intelligence and Machine Learning

What Role Will NLP Play in the Future?

There's no question that natural language processing will play a prominent role in future business and personal interactions. Personal assistants, chatbots and other tools will continue to advance. This will likely translate into systems that understand more complex language patterns and deliver automated but accurate technical support or instructions for assembling or repairing a product.

NLP will also lead to more advanced analysis of medical data. For example, a doctor might input patient symptoms and a database using NLP would cross-check them with the latest medical literature. Or a consumer might visit a travel site and say where she wants to go on vacation and what she wants to do. The site would then deliver highly customized suggestions and recommendations, based on data from past trips and saved preferences.

For now, business leaders should follow the natural language processing space—and continue to explore how the technology can improve products, tools, systems and services. The ability for humans to interact with machines on their own terms simplifies many tasks. It also adds value to business relationships.

Also see: The Future of Artificial Intelligence

Natural Language ProcessingCollege Of Liberal ArtsRIT

Alm, Cecilia O., et al. "Achieving diversity in AI-focused graduate research traineeships." Proceedings of the 55th ACM Technical Symposium on Computer Science Education, V2. Ed. Bell Stephenson, et al. Portland, OR: ACM, 2024. Web.

Alm, Cecilia O. "Centering humans in artificial intelligence." Proceedings of the AAAI 2024 Spring Symposium Series, Vol. 3, No. 1. Ed. Ron Petrick and Christopher Geib. Stanford, California: AAAI, 2024. Web.

Orr, Hayden, Michael Peechatt, and Cecilia O. Alm. "MULTICOLLAB-ASL: Towards affective computing for the Deaf community." Proceedings of the 26th International ACM SIGACCESS Conference on Computers and Accessibility (Poster/Demo Track). Ed. David Flatla, et al. St. John's, NL, Canada: ACM, 2024. Web.

Tigung, Rajesh and Cecilia O. Alm. "FUSE - FrUstration and Surprise Expressions: A subtle emotional multimodal language corpus." Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation. Ed. Nicoletta Calzolari, et al. Torino, Italia: ACL, 2024. Web.

Peechatt, Michael, Cecilia O. Alm, and Reynold Bailey. "MULTICOLLAB: A multimodal corpus of dialogues for analyzing collaboration and frustration in language." Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation. Ed. Nicoletta Calzolari, et al. Torino, Italy: ACL, 2024. Web.

Kothari, Ammina, et al. "News consumption helps readers identify model-generated news." Proceedings of the Proceedings of the 2023 IEEE Western New York Image and Signal Processing Workshop. Ed. N.A. Rochester, NY, USA: IEEE, 2023. Print.

Amin, Akhter Al, et al. "Modeling word importance in conversational transcripts: Toward improved live captioning for deaf and hard of hearing viewers." Proceedings of the Proceedings of the 20th International Web for All Conference. Ed. N. A. Austin, Texas: ACM, 2023. Web.

Dust, A'di, et al. "Understanding differences in human-robot teaming dynamics between deaf/hard of hearing and hearing individuals." Proceedings of the Companion of the 2023 ACM/IEEE International Conference on Human-Robot Interaction. Ed. Ginevra Castellano, et al. Stockholm, Sweden: ACM, 2023. Web.

Arthur, Isabelle, et al. "MDE – Multimodal data explorer for flexible visualization of multiple data streams." Proceedings of the Proceedings of the Affective Computing and Intelligent Interaction Conference: Late Breaking Results. Ed. N. A. Cambridge, MA, USA: IEEE, 2023. Web.

Baheri, Ali and Cecilia O. Alm. "LLMs-augmented contextual bandit." Proceedings of the FMDM Workshop (at NeurIPS 2023). Ed. Sherry Yang et al. N. P., USA: NeurIPS, 2023. Web.

Ashwath, Anisha, et al. "Early vs. Late multimodal fusion for recognizing confusion in collaborative tasks." Proceedings of the 11th International Conference on Affective Computing and Intelligent Interaction Workshops and Demos (ACIIW). Ed. N. A. Cambridge, MA, USA: IEEE, 2023. Web.

Alm, Cecilia O. And Alex Hedges. "Visualizing NLP in Undergraduate Students' Learning about Natural Language." Proceedings of the Proceedings of the AAAI Conference on Artificial Intelligence (EAAI). Ed. Association for the Advancement of Artificial Intelligence. Virtual, Virtual: AAAI Press, 2021. Web.

Hassan, Saad, Matt Huenerfauth, and Cecilia O. Alm. "Unpacking the Interdependent Systems of Discrimination: Ableist Bias in NLP Systems through an Intersectional Lens." Proceedings of the Findings of the Association for Computational Linguistics: EMNLP 2021. Ed. Xuanjing Huang, Lucia Specia, and Scott Wen-tau Yih. Punta Cana, Dominican Republic: Association for Computational Linguistics, 2021. Web.

Kaushik, Nikhil, et al. "Eliciting Confusion in Online Conversational Tasks." Proceedings of the Proceedings of the 9th International Conference on Affective Computing and Intelligent Interaction Workshops and Demos (at ACII 2021). Ed. NA. Nara, Japan (Online): IEEE, 2021. Web.

Akhbardeh, Farhad, et al. "Handling Extreme Class Imbalance in Technical Logbook Datasets." Proceedings of the Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Long Papers). Ed. Chengqing Zong, et al. Online, Online: Association for Computational Linguistics, 2021. Web.

Maus, Natalie, et al. "Gaze-guided magnification for individuals with vision impairments." Proceedings of the Proceedings of Extended Abstracts of the 2020 CHI Conference on Human Factors in Computing Systems. Ed. CHI (Late Breaking Works). Honolulu, HI, USA: ACM, 2020. Web.

Forman, Cleo, et al. "Capturing laughter and smiles under genuine amusement vs. Negative emotion." Proceedings of the Proceedings of the Second Workshop on Human-Centered Computational Sensing (at PerCom 2020). Ed. Amy L. Murphy and Edison Thomaz. Austin, TX, USA [Virtual]: IEEE, 2020. Web.

Kvist, Jonathan, et al. "Dynamic visualization system for gaze and dialogue data." Proceedings of the Proceedings of the Fourth International Conference on Human Computer Interaction Theory and Applications HUCAPP 2020. Ed. Manuela Chessa, Alexis Paljic, Jose Braz. Valetta, Malta: SciTePress, 2020. Web.

Bohlin, Gustaf, et al. "Considerations for Face-based Data Estimates: Affect Reactions to Videos." Proceedings of the Third International Conference on Human Computer Interaction Theory and Applications. Ed. Manuela Chessa, Alexis Paljic, and Jose Braz. Prague, Czech Republic: SCITEPRESS, 2019. Web.

Kafle, Sushant, Cecilia O. Alm, and Matt Huenerfauth. "Modeling Acoustic-prosodic Cues for Word Importance Prediction in Spoken Dialogues." Proceedings of the 8th Workshop on Speech and Language Processing for Assistive Technologies, SLPAT 2019 (at NAACL 2019). Ed. Manuela Chessa, Alexis Paljic, and Jose Braz. Minnesota, MS: ACL, 2019. Web.

Kessler, Lucas, Cecilia O. Alm, and Reynold Bailey. "Synthesized Spoken Names: Biases Impacting Perception." Proceedings of the Interspeech (Show and Tell). Ed. Pejman Mowlaee, Mario Huemer, and Philipp Salletmayr. Graz, Austria: n.P., 2019. Web.

Kafle, Sushant, Cecilia O. Alm, and Matt Huenerfauth. "Fusion Strategy for Prosodic and Lexical Representations of Word Importance." Proceedings of the Interspeech. Ed. Thomas Hain and Björn Schuller. Graz, Austria: ISCA, 2019. Web.

Saraf, Monali, et al. "Multimodal Anticipated Versus Actual Perceptual Reactions." Proceedings of the ICMI19: Adjunct of the 21st ACM International Conference on Multimodal Interaction, Suzhou, China). Ed. Wen Gao, et al. Suzhou, China: ACM, 2019. Web.

Wang, Regina, et al. "Fusing Dialogue and Gaze from Discussions of 2D and 3D Scenes." Proceedings of the ICM19: Adjunct of the 21st ACM International Conference on Multimodal Interaction. Ed. Wen Gao, et al. Suzhou, China: ACM, 2019. Web.

Lucas, Elizabeth, Cecilia O. Alm, and Reynold Bailey. "Understanding Human and Predictive Moderation of Online Science Discourse." Proceedings of the 2019 IEEE Western New York Image and Signal Processing Workshop (WNYISPW). Ed. Raymond Ptucha. Rochester, New York: IEEE, 2019. Web.

Haduong, Nikita, et al. "Multimodal Alignment for Affective Content." Proceedings of the Workshop of Affective Content Analysis (at AAAI 2018). Ed. Niyati Chhaya, Kokil Jaidka, and Lyle H. Ungar. New Orleans, North America: Association for the Advancement of Artificial Intelligence, 2018. Web.

Medina, Rebecca, et al. "Sensing Behaviors of Students in Online vs. Face-to-face Lecturing Contexts." Proceedings of the Workshop on Homan-Centered Computational Sensing (at PerCom 2018). Ed. Cecilia O. Alm, Reynold Bailey, and Ehsan Hoque. Athens, Greece: 2018 IEEE International Conference on Pervasive Computing and Communications Workshops, 2018. Web.

Diaz, Yancarlos, et al. "Towards an Affective Video Recommendation System." Proceedings of the Workshop on Homan-Centered Computational Sensing (at PerCom 2018). Ed. Cecilia O. Alm, Reynold Bailey, and Ehsan Hoque. Athens, Greece: 2018 IEEE International Conference on Pervasive Computing and Communications Workshops, 2018. Web.

Tornblad, McKenna K., et al. "Sensing and Learning Human Annotators Engaged in Narrative Sensemaking." Proceedings of the Student Research Workshop (at NAACL 2018). Ed. Silvio Ricardo Cordeiro, Shereen Oraby, Umashanthi Pavalanathan, and Kyeongmin Rim. New Orleans, Louisiana: Association for Computational Linguistics, 2018. Web.

Meyers, Benjamin S, et al. "A Dataset for Identifying Actionable Feedback in Collaborative Software Development." Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers). Ed. Iryna Gurevych and Yusuke Miyao. Melbourne, Australia: Association for Computational Linguistics, 2018. Web.

Vaidyanathan, Preethi, et al. "SNAG: Spoken Narratives and Gaze Dataset." Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers). Ed. Iryna Gurevych and Yusuke Miyao. Melbourne, Australia: Association for Computational Linguistics, 2018. Web.

Shea, Jordan E., Cecilia O. Alm, and Reynold Bailey. "Contemporary Multimodal Data Collection Methodology for Reliable Inference of Authentic Surprise." Proceedings of the 2018 IEEE Western New York Image and Signal Processing Workshop (WNYISPW). Ed. Ray Ptucha. Rochester, New York: IEEE, 2018. Web.

Gangji, Aliya, Trevor Walden, Preethi Vaidyanathan, Emily Prud'hommeaux, Reynold Bailey, and Cecilia O. Alm. "Using co-captured face, gaze and verbal reactions to images of varying emotional content for analysis and semantic alignment." Proceedings of the Workshop on Human-Aware Artificial Intelligence (at AAAI 2017). Ed. Kartik Talamadupula, et al. San Francisco, CA: AAAI, 2017. Web.

Edwards, Ashley, Anthony Massicci, Srinivas Sridharan, Joe Geigel, Linwei Wang, Reynold Bailey, and Cecilia O. Alm. "Sensor-based methodological observations for studying online learning." Proceedings of the ACM Workshop on Intelligent Interfaces for Ubiquitous and Smart Learning (at IUI 2017). Ed. Ilknur Celik and Ilaria Torre. Limassol, Cyprus, Cyprus: ACM, 2017. Web.

Calderwood, Alexander, Elizabeth A. Pruett, Raymond Ptucha, Christopher M. Homan, Cecilia O. Alm. "Understanding the semantics of narratives of interpersonal violence through reader annotations and physiological reactions." Proceedings of the Workshop on Computational Semantics beyond Events and Roles (at EACL 2017). Ed. Eduardo Blanco, Roser Morante, and Roser Saurí­. Valencia, Spain: ACL, 2017. Web.

Munaiah, Nuthan, Benjamin S. Meyers, Cecilia O. Alm, Andrew Meneely, Pradeep K. Murukannaiah. Emily Prud'hommeaux, Josephine Wolff, and Yang Yu. "Natural language insights from code reviews that missed a vulnerability: A large scale study of Chromium." Proceedings of the International Symposium on Engineering Secure Software and Systems. Ed. Eric Bodden, Mathias Payer, Elias Athanasopoulos. Bonn, Germany: n.P., 2017. Web.

Alm, Cecilia O., Benjamin S. Meyers, and Emily Prud'hommeaux. "An analysis and visualization tool for case study learning of linguistics concepts." Proceedings of the Conference on Empirical Methods in Natural Language Processing (Demo papers). Ed. Lucia Specia, Matt Post, Michael Paul. Copenhagen, Denmark: ACL, 2017. Web.

Alm, Cecilia O. And Reynold Bailey. "Team-based, transdisciplinary, and inclusive practices for undergraduate research." Proceedings of the IEEE Frontiers in Education Conference. Ed. N.P. Indianapolis, IN: IEEE, 2017. Web.

Terkik, Andamlak, Emily Prud'hommeaux, Cecilia O. Alm, Christopher Homan, Scott Franklin,. "Analyzing Gender Bias in Student Evaluations." Proceedings of the COLING, the 26th International Conference on Computational Linguistics: Technical Papers. Ed. Nicoletta Calzolari et al. Osaka, Japan: n.P., 2016. Web.

Liu, Tong, Christopher Homan, Cecilia O. Alm, Megan Lytle, Ann Marie White, Henry Kautz,. "Understanding Discourse on Work and Job-related Well-being in Public Social Media." Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics. Ed. Antal van den Bosch et al. Berlin, Germany: n.P., 2016. Web.

Bullard, Joseph, Cecilia O. Alm, Xumin Liu, Ruben A. Proano, Qi Yu, . "Towards Early Dementia Detection: Fusing Linguistic and Non-linguistic Clinical Data." Proceedings of the Computational Linguistics and Clinical Psychology: From Linguistic Signal to Clinical Reality (at NAACL 2016). Ed. Kristy Hollingshead and Lyle Ungar. San Diego, USA: n.P., 2016. Web.

Oak, Mayuresh, Anil K. Behera, Titus Thomas, Cecilia O. Alm, Emily Prud'hommeaux, Christopher M. Homan, Ray Ptucha,. "Generating Clinically Relevant Texts: A Case Study on Life-changing Events." Proceedings of the Computational Linguistics and Clinical Psychology: From Linguistic Signal to Clinical Reality (at NAACL-HLT 2016). Ed. Kristy Hollingshead and Lyle Ungar. San Diego, California: n.P., 2016. Web.

Vaidyanathan, Preethi, Jeff B. Pelz, Emily Prud'hommeaux, Cecilia O. Alm, Anne R. Haake,. "Fusing Eye Movements and Observer Narratives for Expert-driven Image-region Annotations." Proceedings of the ETRA. Ed. Pernilla Qvarfordt and Dan Witzner Hansen. Charleston, SC: n.P., 2016. Web.

Schrading, Nicholas, et al. "An Analysis of Domestic Abuse Discourse on Reddit." Proceedings of the Conference on Empirical Methods in Natural Language Processing, pages 2577-2583. Ed. Lluís Màrquez, Chris Callison-Burch, Jian Su. Lisbon, Portugal: ACL, 2015. Web.

Vaidyanathan,, Preethi, et al. "Computational Integration of Human Vision and Natural Language Through Bitext Alignment." Proceedings of the Workshop on Vision and Language at the Conference on Empirical Methods in Natural Language Processing, pages 4-5. Ed. --. Lisbon, Portugal: ACL, 2015. Web.

Schrading, Nicholas, et al. "#WhyIStayed, #WhyILeft: Microblogging to Make Sense of Domestic Abuse." Proceedings of the 2015 Conference of the North American Chapter of the Association for Computational Linguistics-Human Language Technologies, pages 1281-1286. Ed. --. Denver, CO: ACL, 2015. Web.

Bethamcherla, Vasudev, et al. "Face-speech Sensor Fusion for Non-invasive Stress Detection." Proceedings of the 1st Joint Conference on Facial Analysis, Animation, and Auditory-Visual Speech Processing, pages 196-201. Ed. --. Vienna, Austria: ISCA, 2015. Web.

Paul, Will, et al. "Stressed out: What Speech Tells us About Stress." Proceedings of the Interspeech 2015, pages 3710-3714. Ed. --. Dresden, Germany: ISCA, 2015. Web.

Vaidyanathan, Preethi, et al. "Alignment of Eye Movements and Spoken Language for Semantic Image Understanding." Proceedings of the 11th International Conference on Computational Semantics, pages 76-81. Ed. --. London, UK: ACL, 2015. Web.

Bullard, Joseph, et al. "Inference from Structured and Unstructured Electronic Medical Data for Dementia Detection." Proceedings of the Operation Research and Computing: Algorithms and Software for Analytics, 14th INFORMS Computing Society Conference (ICS2015), pages 236—244. Ed. --. Richmond, VA: n.P., 2015. Web.

Guo, Xuan, et al. "Fusing Multimodal Human Expert Data Towards Semantic Image Use." Proceedings of the 7th Workshop on Eye Gaze in Intelligent Human Machine Interaction: Eye-Gaze & Multimodality. Ed. Nb. Istanbul, Turkey: ACM, 2014. Web.

Hochberg, Limor, et al. "Towards Automatic Annotation of Clinical Decision-making Style." Proceedings of the LAW VIII - The 8th Linguistic Annotation Workshop at COLING 2014. Ed. Lori Levin, Manfred Stede. Dublin, Ireland: n.P., 2014. Web.

Bullard, Joseph, et al. "Towards Multimodal Modeling of Physicians Diagnostic Confidence and Self-awareness Using Medical Narratives." Proceedings of the COLING 2014, the 25th International Conference on Computational Linguistics: Technical Papers. Ed. Junichi Tsujii and Jan Hajic. Dublin, Ireland: Dublin City University and Association for Computational Linguistics, 2014. Web.

Moseley, Nathaniel, Cecilia O. Alm, and Manjeet Rege. "Toward Inferring the Age of Twitter Users From Their Use of Nonstandard Abbreviations and Lexicon." Proceedings of the IEEE IRI 2014. Ed. Elisa Bertino, Bhavani Thuraisingham, Ling Liu, James Joshi. San Francisco,, California: IEEE, 2014. Web.

Hochberg, Limor, Cecilia O. Alm, Esa M. Rantanen, Caroline M. DeLong, and Anne R. Haake. "Decision Style in a Clinical Reasoning Corpus." Proceedings of the 2014 Workshop on Biomedical Natural Language Processing (BioNLP 2014),. Ed. Kevin Bretonnel Cohen et al. Baltimore, Maryland: ACL, Web.

Christopher, Homan, et al. "Toward Macro-Insights for Suicide Prevention: Analyzing Fine-Grained Distress at Scale." Proceedings of the Workshop on Computational Linguistics and Clinical Psychology: From Linguistic Signal to Clinical Reality at the 52nd Annual Meeting of the Association for Computational Linguistics. Ed. Resnik, P., Resnik, R., Mitchell, M. Baltimore, Maryland: n.P., Web.

Moseley, Nathaniel, Cecilia O. Alm, and Manjeet Rege. "User-annotated Microtext Data for Modeling and Analyzing Sociolinguistic Characteristics and Age Grading." Proceedings of the IEEE Eighth International Conference on Research Challenges in Information Science. Ed. Rebecca Deneckere, Marko Bajec, Martine Collard. Marrakesh, Morocco: IEEE, Web.

Guo, Xuan, et al. "Infusing Perceptual Expertise and Domain Knowledge into a Human-centered Image Retrieval System: A Prototype Application." Proceedings of the ETRA 2014. Ed. Nb. Safety Harbor,, Florida: ACM, Web.

Vaidyanathan, Preethi, et al. "Recurrence Quantification Analysis Reveals Differences in Eye-movement Behavior Between Expert and Novice Dermatologists." Proceedings of the ETRA 2014. Ed. Unknown. Safety Harbor, Florida: n.P., 2014. Web.

Womack, K., et al. "Markers of Confidence and Correctness in Spoken Medical Narratives." Proceedings of the Interspeech 2013. Ed. ISCA. Lyon, France: n.P., 2013. Web.

Womack, Kathryn, et al. "Using Linguistic Analysis to Characterize Conceptual Units of Thought in Spoken Medical Narratives." Proceedings of the Interspeech 2013. Ed. ISCA. Lyon, France: n.P., 2013. Web.

Womack, Kathryn, et al. "Disfluencies as Extra-Propositional Indicators of Cognitive Processing." Proceedings of the Ex-Prom-2012: Workshop on Extra-Propositional Aspect of Meaning in Computational Linguistics at the 50th Annual Meeting of the Association for Computational Linguistics 2012. Ed. Roser Morante and Caroline Sporleder. Jeju, Korea: n.P., 2013. Web.

McCoy, Wilson. "Annotation Schemes to Encode Domain Knowledge in Medical Narratives." Proceedings of the 6th Linguistic Annotation Workshop at the 50th Annual Meeting of the Associate. For Computational Linguistics 2012. Ed. Nancy Ide and Fei Xia. Jeju, Korea: n.P., 2012. Web.

McCoy, Wilson, et al. "Linking Uncertainty in Physicians' Narratives to Diagnostic Correctness." Proceedings of the ExProm-2012: Workshop on Extra-Propositional Aspect of Meaning in Computational Linguistics at the 50th Annual Meeting of the Association for Computational Linguistics. Ed. Roser Morante and Caroline Sporleder. Jeju, Korea: n.P., 2013. Web.

Lehrman, Michael T., Cecilia O. Alm, and Ruben A. Proano. "Detecting Distressed vs. Non- Distressed Affect State in Short Forum Texts." Proceedings of the 2012 Workshop on Language in Social Media at the Conference of the North Am. Chapter of the Associate. For Comp. Linguistics-Human Language Technologies. Ed. ACL/NAACL-HLT (Meena Nagarajan, Sara Owsley Sood, Michael Gamon). Montreal, Canada: n.P., 2012. Web.

Li, Rui, et al. "Learning Eye Movement Patterns for Characterization of Perceptual Expertise." Proceedings of the Symposium on Eye Tracking Research and Applications (ETRA). Ed. ETRA. Santa Barbara, CA: n.P., 2012. Web.

Alm, Cecilia O. "Subjective Natural Language Problems: Motivations, Applications, Characterizations, and Implications." Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies. Ed. Dekang Lin, Yuji Matsumoto, Rada Mihalcea. Portland, OR: The Association for Computational Linguistic, 2011. Web.

What Are Some Controversies Surrounding Natural Language Processing?

NEWYou can now listen to Fox News articles!

As machine learning technology continues to shock the world, popular artificial intelligence tools such as natural language processing may generate unforeseen issues for humanity. 

For instance, natural language processing can have implicit biases, create a significant carbon footprint, and stoke concerns about AI sentience. Natural language processing is a field in machine learning where a computer processes human language through vast amounts of data to understand, translate, extract, and organize information. However, the language processing tools such as Open AI's Chat GPT and other tools run into some challenges, such as misspellings, speech recognition, and the ability of a computer to understand the nuances of human language. 

One of the biggest rising concerns regarding natural language processing is artificial intelligence programs' ability to have implicit bias and perpetuate stereotypes. One of the most essential tasks of natural language learning models is to study and learn patterns from data sets in order to understand how humans communicate with one another. Sometimes, these data sets can have implicit bias thinking that may affect how an AI learns the language and communicates its findings. 

WHAT IS THE HISTORY OF AI?

For example, suppose a dataset has language that assigns certain roles to men, such as computer programmers or doctors but assigns roles, like homemaker or nurse, to women. In that case, the AI program will implicitly apply those terms to men and women when communicating in real time. Therefore, stereotypes existing within the data set can lead to algorithms having language that applies unfair stereotypes based on race, gender, and sexual preference. 

Popular natural language processing programs such as Open AI's ChatGPT has had issues with stereotypes and political bias in the past.  (iStock)

Political bias is another real concern for natural language processing programs that may lead to the impression of information based on the political preference of the data set used to train the AI. For instance, in February 2023, ChatGPT users discovered that the language processing program refused to communicate information about the Hunter Biden laptop story and speak about former President Donald Trump positively despite doing the same for President Joe Biden. 

The political biases of machine learning language processing tools often result directly from the programmer or the dataset it is trained with. If the programmer refuses to correct those biases, it often leads to the suppression of news and information that may anger one side of the political spectrum. 

WHAT ARE THE FOUR MAIN TYPES OF ARTIFICIAL INTELLIGENCE? FIND OUT HOW FUTURE AI PROGRAMS CAN CHANGE THE WORLD

Read below to discover other controversies and concerns regarding natural language processing. 

Coherence versus sentience

One concern that individuals have had about the AI industry for years is a machine learning programs' ability to seemingly think for themselves and express feelings. Natural language processing models are often the version of AI that concerns individuals in this regard due to the computer's ability to mimic and present written text in a way that expresses the same emotions and thought patterns as humans.

Artificial Intelligence does not have the ability to express feelings or think for themselves in a sentient manner.  (REUTERS/Dado Ruvic/Illustration)

However, just because an AI program is coherent or as the ability to readily generate information does not mean the machine is sentient. It is not possible for AI to register experiences or feelings because it does not have the ability to think, feel, or perceive the world with a sentient mind. 

Environmental impact

Artificial intelligence, in general, but specifically natural language processing models, creates an environmental footprint that is comparable to the oil industry. Data mining, which is essential for the existence of artificial intelligence, consumes a large amount of electricity which releases carbon dioxide into the air. For instance, the data mining generated from cryptocurrency and AI-related programs between 2021-22 was responsible for an excess of 27.4 million tons of carbon dioxide into the air. 

WHAT ARE THE DANGERS OF AI? FIND OUT WHY PEOPLE ARE AFRAID OF ARTIFICIAL INTELLIGENCE

Natural language processing is a lucrative commodity yet has one of the largest environmental impacts out of all the other fields in the artificial intelligence realm. The process used to train, experiment, and fine-tune a natural language process model has been estimated to create on average more CO2 emissions than two Americans annually. 

Data mining related to artificial intelligence has had a significant environment impact via CO2 emissions.   (JOSEP LAGO/AFP via Getty Images)

CLICK HERE TO GET THE FOX NEWS APP

Some natural language processing programs that use neural architecture search created even more CO2 emissions that experts have estimated to be nearly five times more than the carbon footprint of a normal American car driver. 

---