Based in Sydney, Australia this a Blog of Associate Professor AmandeeP Sidhu. it focuses on everything Artificial Intelligence, Big Data and Biomedical Informatics.

Artificial Intelligence Action in 2018

I agree with my friend Bernard Marr that Artificial Intelligence (AI) Hype of 2017 will change to Artificial Intelligence Action in 2018. Gartner predicts that AI will create 2.3 million jobs in 2020, while eliminating 1.8 million.

AI in 2017 has been used as a loose umbrella term for technologies: Machine Learning (ML) which includes Deep Learning (DL), Natural Language Processing (NLP), Computer Vision (CV), Automated Reasoning (AR), and Artificial Generalized Intelligence (AGI) or Strong AI. Most of the AI Applications in 2017 and 2018 will be Weak AI – focused on one narrow task. All ML(DL), NLP, CV, AR tasks irrespective of how they are marketed are Weak AI.

AGI or Strong AI attempts to simulate general human thought processes by using a computerized model of concepts to organize knowledge and then act on it. Instead of being programmable like Weak AI, Strong AI seeks to make sense of the world by relying on human language’s inherent model of reality by using logic. Semantic Web can finally be of some help here.

Most commercial applications in 2017 consisted of multiple types of AI applied or configured in conjunction with one another and other technologies. For example, language translations are supported by a combination of NLP and DL; autonomous machines use localization and mapping powered by ML, CV, and sensors. In 2018, Strong AI will be more centre stage and Weak AI will be more integrated in our daily lives.

In 2018 AI will not be the only show in town; numerous other technologies will both leverage and influence AI’s development, adoption, and regulation:

  • Internet of Things (IoT)

  • Blockchain

  • Cryptography and Cybersecurity

When developing various AI solutions in 2018 we will face a range of practical limitations and issues like:

  • Access to quality training Data

  • Data cleaning and processing, ensuring Data Integrity

  • Lack of interoperability across protocol, device types, data types, and data sets

  • Constraints in “edge-level” processing power

  • Lack of algorithmic standards

  • Risks of algorithmic explainability

  • Unclear ethical standards although EU is doing some initial work

  • Regulatory questions & Compliance with existing regulations

Artificial Intelligence Statistics by Sector

Artificial Intelligence Statistics by Sector

These will influence AI adoption in 2018. While certain jobs will become automated, AI is more poised to augment humans in 2018. We are already starting to see that many ethical, philosophical, cultural, societal, and business norms will be forced into re-assessment in 2018.

It's all about Data