AIArtificial IntelligenceblogData Science

Conversational AI and the Future of Human-Technology Interaction

Conversational AI is the process of using computer systems to allow us to be more efficient at communication.  Computer systems have been able to convert  text-to-speech and speech-to-text for some time now but with the advent of powerful AI models, chatbots were developed and are quickly evolving into seamless human to computer speech interactions..  

These systems show great promise when it comes to making our business and personal lives easier. In the 1990’s the internet started to provide a huge leap in terms of our ability to gather information and learn autonomously.  New conversational AI solutions, and the underlying large language models (LLMs), will make that process even more seamless.   And the internet is a primary enabler in this next evolution as the vast amounts of data that is now stored on the internet are inputted into these AI solutions so we as individuals may now find most of what we need quickly.    

Knowledge Versus Reasoning

With current supercomputers and cloud-based computer systems, and the extensive amount of digital data now available on nearly every topic imaginable, computer systems can now digest knowledge at a rate that was difficult to conceive even a decade ago.  And all that data / knowledge is used as inputs into AI models and it is the AI models that generate the reasoning which allows for more advanced computer solutions such as Conversational AI, a subset of Generative AI, and full self- driving.   AI systems can not mimic the level of human reasoning but the advances over the past five to ten years has enabled the ability for computers to have a conversation with a human.  It is still not near the level of human-to-human conversations but for the purpose of getting information quickly in a way that is understandable we are already there.      

Conversational AI Solutions

Many AI-based Conversational AI solutions: are available which include: 

  • Speech-to-Text (STT):    Speech-to-text or automatic speech recognition (ASR) translates human spoken language into written text.   STT technology has been around for years but the advent of powerful AI models has made the output much more precise and the process of translating speech to text easier. Amazon Transcribe is one solution used by businesses and free consumer solutions are available like Dictanote.     
  • Text-to-Speech (TTS):   Google’s NotebookLM now let’’s you listen to a conversational podcast based on your inputted content (text).  This is currently a free web-based solution and the input content can be an article, webpage, Google Doc or any other text-based document.   OpenAI has a TTS module you may use programmatically as well to generate text-to-speech.
  • Speech-to-Speech (STS): STS is the process of asking the AI system questions and the system responding.  These virtual assistant type systems include Apple’s Siri and Amazon’s Alexa.   Larger organizations are building their own internal AI-driven chatbots and solutions are available such as Zendesk and Yellow AI.  Full conversational STS AI systems are now becoming public including Flow and the the ChatGPT App.    

Conversational AI: Will Computers become self aware?

If you were to input all digital data available into one computer system could that computer learn, reason and behave like a human?  Not now.  AI is not at a level where it can continually self-learn on its own.  Self learning techniques are available but an AI system can’t yet take control, and continue to become smarter and reason autonomously.  Computers and AI systems are not at a level that allows them to understand and identify how to build on their own to solve a problem or reach a goal.      

Could AI systems evolve on their own?  For specific tasks or functions possibly but they need to first learn how to learn more efficiently – which is happening now.  Artificial General Intelligence (AGI) is where a computer system has cognitive skills like a human in a variety of domains.  Many of the experts believe that we will eventually reach AGI.  It is not a question of if but when.  

Could AI systems become self aware?  Ten years ago I would have said no chance.  Now, unlikely.  A self-aware being constantly evaluates and makes decisions based on character and has empathy towards others, the earth and beyond.   If decisions are not constantly made with empathy and a conscience, history has shown that the result is a greater chance of turmoil and wars.  .The WOPR computer in the movie War Games got it right but that was a movie.           

Conversational AI Podcast

I recently released an article on the applications of AI in business which covers some of the topics above in greater depth and the podcast embedded below was AI generated from that article.   This is one example of using Conversational AI to gather information more seamlessly as you can listen to a podcast without your full undivided attention which is required when reading an article.  The AI solution that generated the podcast interleaved topics not in the article in order to make the podcast format more enjoyable and seamless.  It was able to do this as the underlying model was built on several massive and varied data sources.