Artificial Intelligence: The Business Applications

Artificial Intelligence: An Overview and Evolving Responsively

Artificial Intelligence (AI) involves creating intelligent computer systems that operate autonomously. These systems possess the ability to learn, adapt, make decisions and act in a manner similar to the human brain, utilizing advanced techniques such as Large Language Models (LLMs) and Generative AI.  AI systems can generate original content (text/language), visualizations, programming code, insights and are the brains within recognition devices and automated mechanical systems like robots and self-driving.          

AI is moving into many facets of our lives and the impact of these systems will continue to escalate.  Will AI eliminate jobs?  Yes in certain cases but it will also open up other career avenues as well as make some positions less repetitive thus allowing individuals to focus on higher value tasks.     

OpenAI helped broadcast the power of AI to the masses by releasing ChatGPT.  A spotlight has also been placed on the need to ensure that AI is harnessed for good and properly implemented and the president of the U.S. recently issued an executive order on the ‘Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence’ .  Additional groups, public and private, will be formed with the sole focus of ensuring the responsible use of AI.    

AI has enormous potential, including assisting with medical breakthroughs,  however AI needs to be monitored and controlled as the systems become more intelligent.  A supercomputer named Hal took over a spacecraft in the movie 2001: A Space Odyssey. This may have been viewed as pure science fiction when the movie was released in 1968 but with advances over the past 55+ years safeguards need to be put into place to ensure that future AI systems do not behave like Hal.   

Incorporating Artificial Intelligence into Business & Customer Interactions 

Companies are using AI to streamline processes, increase profitability and enhance their customer’s experience in several areas including:  

  • Personalized Content:  Advances in AI allow for personalized content and interactions with customers.  This can include content tailored to an individual’s need or interests in the form of automated email and chat communications 
  • Personalized Offers:   AI may be used to determine the next best offer based on known past purchases and interactions including digital behavior.  
  • Increase Profit & Reduce Fraud: There are many AI applications used to increase profit including models used to determine an individual’s credit worthiness and potential value.  AI is also used to identify fraudulent financial transactions. 
  • Customer Identification:  AI is used in visual and voice identification in order to streamline communications and enhance the customer experience.  
  • Product Delivery: AI is used to optimize inventory management and the delivery of products to customers.  AI is also used to enhance and automate communications while deliveries are in transit: 
  • Streamlining System Development:  Streamlining development can impact profit as it allows for the creation and implementation of technical systems quicker through the use of self-learning models and auto code generators. 
  • Enhanced Communications:  Techniques such as sentiment analysis can help better identify disinterest or satisfaction and businesses use those insights in real time to ensure that the proper communication is delivered.   

Artificial Intelligence versus Data Science

AI differs from traditional data science however overlap does exist.  AI develops machines that mimic human intelligence resulting in recommendations, visual & speech recognition, original text, language translations and decision making to name a few while data science is used to create insights, predictions and classifications.  

Generative AI is a specific type of AI which generates original content based on a model built on existing content.  Generative language models generate new text / language or code while generative image models generate new images or video.     

One of the primary differences between AI and data science is that AI is designed to learn and adapt autonomously.  Each can be enhanced over time, with evolving techniques and updated data, however AI models have the ability to self-learn.  This does not mean that AI is better than data science in all applications where overlap exists – at least not yet.  AI and data science are similar in a few areas including: 

  • The power of each is contingent on the data inputs.  If little data is available it is likely that both techniques will not yield meaningful results.   
  • Both can use neural network techniques as the base underlying model or a portion of the final model. 
  • Those who build these systems need to have a deep understanding of computer science, have strong big data experience, and when the solution is built for a specific industry the developers (or data scientists) should also have a solid business acumen in that specific field  

A factor in determining whether to use data science or AI is how the system will be applied.  Typically, AI models are very large and require significant computing power where some data science techniques allow for less computing power and a potential reduction in the data inputs.  Ease of implementation is a key deciding factor in any system development.  

AI: The Solutions

The purpose here is not to provide an exhaustive list of available AI solutions but I touch on a few below that represent a cross section of applications that are used in business analytic and insight environments:

  • Generative AI:  Generative AI uses data and text inputs to generate original content including text (language) and images.  The most well known generative AI solution is OpenAI’s ChatGPT.  There are AI solutions that you can access through an API like ChatGPT and open source AI solutions are available that may be downloaded locally like Falcon AI
  • Code Generators: Related to generative AI are code generation AI models which produce programming code based on required inputs called prompts.  Facebook recently released Code LLama  which is their auto-coding Large Language Model (LLM).  LLama generates code for some of the more popular languages including Python and Java.  Open source code generation LLMs are available as well like StarCoder. SQL specific generators are available including  SQL Chat & AI Query.  
  • Conversational Data Analysis: Packages like PandasAI, a Python library, allows developers to tap into the available LLMs (ChatGPT, Google Palm,,..) through an API, or open source LLMs like Falcon AI, and incorporate those large models into codeless and conversational data analysis.  PandasAI also incorporates other Python libraries to generate conversational and codeless graphs and predictive models.    
  • Natural Language Processing (NLP):  NLP techniques help computers understand human language and use those insights in generative AI and code generators.    NLP is also used in speech recognition. 
  • Neural Networks and Predictive AI: Neural Networks may be used to generate predictions such as the next best targeted product and within social media recommendation engines.   Tensorflow is one of the more popular machine learning libraries which incorporates Neural Networks.     

AI: Final Thoughts

AI shows extreme the area of business analytics.  Yes, AI will eventually replace some jobs but like many innovations it will allow businesses to become more efficient and enable many workers to focus on higher value tasks.  As an example, the conversational SQL generators and data analysis solutions will allow a broader audience to quickly generate insight from data not possible previously and answer questions that typically are sent to a business analytics or developer group, added to a queue, and then delivered when a developer or analyst is available.  These AI solutions will also be used as training tools and help newer technical and analytic employees get up to speed quicker.   

LLM models are typically large but as more open source versions become available, that may be downloaded locally (Falcon AI, StarCoder,..), this will enable business groups to adopt AI quicker as data privacy is less of a concern (compared to using APIs) and using the AI solutions locally will allow teams to become better acquainted with the solutions.   .  

As these AI systems are being developed we will need to make sure Hal does not go off the rails but I am confident in our ability to safely evolve.  AI is coming to a business group near you.