If you’ve been online, read the news, or followed business trends in the past year, you’ve likely encountered discussions about artificial intelligence (AI) and many questions about its limitations and potential. In this article, the Lyssn team aims to provide a foundational understanding of AI for those without a computer science background seeking to make sense of this new technology.
The AI basics
AI is a field of computer science focused on creating systems capable of performing tasks and recognizing patterns that typically require manual and tedious human attention. These tasks include learning, pattern identification, perception, language recognition, and recommending decisions based on past information.
The applications of this type of technology are endless, but it’s important to note that there are different types of AI. Classification and Generative AI models are limited to their given parameters and mainly focus on recognizing patterns or key terms from the training data. Thus, the AI models will only produce answers that are as good as the data they were given.
The three types of AI discussed regularly in tech circles:
Classification AI — designed to perform a specific prediction or pattern recognition on a narrow range of tasks. It operates under a set of constraints and cannot perform tasks outside its designated domain.
Examples: Voice assistants like Siri or Alexa, recommendation systems like those used by Netflix or Amazon, or image recognition systems used in photo tagging
Identifying characteristics:
- Highly specialized
- Limited scope of operation
- Trained to recognize patterns or terms from training data (Data used to teach an AI model, helping it learn to make accurate predictions or decisions.)
- Does not possess consciousness, the ability to reason, or general understanding
Generative AI (GenAI) — a subset of AI focused on generating new content, such as text, images, music, or even code. It uses past information to generate new data that mimics real-world data.
Examples: ChatGPT, which can generate human-like text, DALL-E, which creates images from textual descriptions
Identifying characteristics:
- Can produce new, original content based on training data
- Often employed in creative industries, content generation, and design
- Does not possess consciousness, the ability to reason, or general understanding
General AI (Strong AI or Artificial General Intelligence – AGI)— a theoretical type of AI that is not in use yet in the public mainstream. It refers to an artificial intelligence system with the ability to understand, learn, and apply intelligence across a wide range of tasks, similar to a human being.
Examples: As of now, true General AI does not exist. It remains a theoretical concept and a long-term goal for AI researchers.
Identifying characteristics:
- Versatile and adaptive
- Can perform any intellectual task that a human can
- Possesses the ability to understand, reason, and learn broadly and deeply
- Capable of transferring knowledge from one domain to another
What type of AI does Lyssn use?
Lyssn uses Classification AI across its entire platform, except for the optional note documentation feature within our QI platform, which uses a Generative AI model highly-specialized for health and human services sessions. To learn more about Lyssn’s work with various types of AI, take a look at a statement from our CEO
David Atkins
, or download Generative AI Quick Fact Sheet.
How do I know if a tech company is actually using AI?
We’re glad you asked! Our Chief Science Officer,
Zac Imel
, provides a simple list of 6 questions you can ask to evaluate any company in a blog post & white paper “Is that AI…Really AI?”
What’s different about Lyssn’s AI compared to other AI companies?
Importantly, Lyssn’s proprietary technology is not built on top of ChatGPT or any other generative AI application. We utilize 8+years of proprietary data from mental health, child welfare, wellness, and social services interactions as the foundation for informing our AI. So far, Lyssn has individually labeled 4B words from 26K+ psychotherapy conversations. Our AI team consists of licensed mental health professionals who manually code/annotate the de-identified conversations that are then used to train our AI models for evidence based practices.
Translation of AI jargon into layman’s terms
- Algorithm: The specified set of rules that a machine will follow to learn how to do a task & then perform that same task over and over again. Algorithms are unique to each company and program.
- Automation: When AI technology performs tasks and processes without human intervention. Depending on the AI, these tasks can range from simple & repetitive to more complex, decision-making tasks.
- Bias: The discrimination that can occur in AI systems, which can lead to unfair decisions, responses, or outcomes that disproportionately affect certain groups of individuals. The source of bias in AI is often a result of lack of diversity in the data set the AI was trained on or biases introduced by the algorithms themselves. Lyssn does a yearly evaluation of the bias in our system, you can see our 2023 report here.
- Big Data: An extremely large and complex dataset. This is often very difficult for traditional data processing tools to analyze due to the volume, variety, or sheer speed that the data is being collected. models.
- Chatbot: A?messaging program designed to simulate conversation with human users, often used to provide information or support in real-time.
- Classification AI: AI systems designed to perform a specific task or set of tasks, such as recognizing images or speech, rather than general problem-solving.
- Data Coding: The process of organizing and categorizing data in a systematic way, making it easier for an AI model to analyze and identify.
- Data Labeling: The task of tagging or annotating data with labels that provide meaningful information, crucial for training AI models.
- Data Mining: The practice of examining large datasets to discover patterns, trends, or useful information for a more specific purpose.
- Dataset: A collection of related data points or information, usually organized in a structured format, used for analysis or training AI.
- Deep Learning: A type of machine learning that uses complex algorithms and neural networks to analyze data and make decisions, mimicking the way the human brain works.
- GenAI (Generative AI): A branch of AI that can create new content, such as text, images, or music, based on patterns learned from existing data.
- General AI: An advanced form of AI that can perform any intellectual task a human can, with the ability to learn and adapt to different situations.
- Hallucination (in the context of AI): When an AI system generates information or answers that seem real but are actually incorrect or nonsensical.
- Human Annotators: Individuals who review data (calls, transcripts, or other forms of data) and manually label the data before the AI is trained on it.
- Human Coders: Individuals who write and develop computer programs, creating the instructions that tell computers and AI systems what to do.
- LLMs (Large Language Models): Highly advanced AI systems trained on vast amounts of text data, capable of recognizing or generating human-like language.
- Machine Learning: A method of teaching computers to learn from and make decisions based on data, improving their performance over time without being explicitly programmed.
- Machine Translation: The use of AI to automatically translate text or speech from one language to another.
- Natural Language Processing (NLP): A field of AI focused on enabling computers to identify, interpret, and respond to human language in a meaningful way.
- Neural Networks: A series of algorithms modeled after the human brain, used in machine learning to recognize patterns and solve complex problems.
- Program: A set of instructions written by humans that tell a computer how to perform a specific task.
- Platform: A software or hardware environment where applications run, providing the tools and resources needed for development and operation.
- Strong AI / General AI / AGI: A theoretical type of AI that possesses the ability to understand, learn, and apply knowledge in a way that is indistinguishable from human intelligence.
- Training Data: The data used to teach an AI model, helping it learn to make accurate predictions or decisions.
- Variation / Utterance: Different ways of expressing the same idea or information, often used in training AI to recognize diverse language inputs.
Looking to learn more about our Lyssn products for HHS? Reach out to our team for more information.
