Thoughtware

Article 11/09/2018

Artificial Intelligence for R&D Tax Credits

Pillars on a Government building

Opportunities exist for research credits when applying computer systems to tasks that once required human intelligence. Time spent training artificial intelligence (AI) systems to use data and guiding the system through making distinctions could qualify for the research and development (R&D) tax credit. AI today isn’t always virtual helpers like Siri or Alexa. Companies will operate with various levels of AI integration:

  • Automation of tasks using script or macros
  • Application of automation tools tactically to meet distinct goals
  • Range of automation tools to apply AI to multiple processes
  • Application of AI and automation throughout its operations

Several factors have created the opportunity for research and development of AI, including big data, social media and the internet and cheap storage. The following are the top startup industries for AI:

  • Cybersecurity
  • Health Care
  • Enterprise
  • Fintech/Insurance
  • Marketing/Sales/CRM

AI has eight core capabilities. The first four capabilities focus on capturing information, while the second four focus on figuring out what’s happening:

  1. Speech and recognition – Allow direct commands
  2. Image recognition – Tagging images and making distinctions among them
  3. Search
  4. Clustering
  5. Natural language understanding (NLU) – Beyond voice recognition; requires degree of understanding and cognition
  6. Optimization – Transforms data from one form to another. Most often requires applying algorithms and cognitive reasoning to solve problems
  7. Prediction – Uses historical data to assess new data. For example, making restaurant recommendations, analyzing spending habits or analyzing risk factors in a loan application
  8. Understanding – Not commercially available yet; would involve the machine’s ability to be consciously aware of what it does or thinks

Areas of Opportunity for AI Capabilities

Wages

  • Designing, developing and testing product prototypes
  • Architecture development of deep neural networks (DNN)
  • Developer training of network connection by weighting
  • Developing and refining algorithms
  • Data modeling
  • Evaluating cognitive robotics creating new levels of judgment-based automation; enabling new levels of productivity and accuracy
  • Robotic process automation (RPA) of repetitive transactional work
  • Programming software source code
  • Establishing electronic interfaces and functional relationships between various software modules
  • Creating or improving virtual agents/representatives that automate dialogue with users in natural languages and understand and resolve complex user queries by applying “deep learning,” such as the ability to automate judgment on images
  • Exploring better sight and sound recognition
  • Testing the use of natural language processing, machine learning and rules to automate
  • Testing the application of AI in robotic controllers

Contract Research

  • Cloud computing server rental expenses for prototyping, experimentation and testing

The federal credit for increasing research activities is an income tax credit that can provide a dollar-for-dollar reduction in a company’s federal income tax liability. For qualified small businesses without any income tax liability, the credits can be used for payroll tax relief if the following criteria apply:

  • Entity is a corporation, partnership or individual
  • Tax year gross receipts are $5 million or less
  • No gross receipts prior to five years preceding the current/claim tax year

For more information on R&D tax credits, contact Jason or your trusted BKD advisor.