Below the buzz, underlying the hype will be some of the concrete real world let’s really use #RealAI to solve real world problems, versus the #ToyAI phenomenon.
1. #AugmentedIntelligence will appear to be more prominent than “artificial intelligence”. The emphasis will primarily be on how to augment human in the loop activities rather than replacing humans. Augmentation will come in the form of automation of the road and repetitive activities such as those prescribed an intelligent automation.
2. #AIEthics of Training AI systems will come to the forefront of controversies and bnegin to resolve themselves. Training on bogus data will produce a bogus artificial neural network. Therefore the veracity and traceability and governance around where we have obtained the training sets the validation sets in the testing sets for machine learning and deep learning will come to the forefront as key issues for artificial intelligence governance.
3. #AIStandards for interoperability between AI vendors, intermediate models of knowledge and training representation and output will dominate the issues clients will have in using AI systems vs AI in the abstract. It’s the wild West right now with most companies having unique formats and representations for inputs and outputs and models. Such as they did in the air of middleware databases application programming interfaces that they need to remain vendor neutral. Most AI systems or application programming interfaces are somewhat vendor dependent even if they’re open sourced.Transitional stages in the machine learning lifecycle as well as subsets of knowledge sets will be used as handoff points between points in the lifecycle for machine learning and artificial intelligence will be increasingly in demand. Vendors will need to provide persistent representation of knowledge set outputs among their own tools and APIs as well as between vendors.
4. #InfuseIntelligence. Augmented Intelligent Systems will not appear as an ‘n’-th state of maturity, but will be infused into the spectrum of legacy to more modern (read “immediate”) AI and ML applications. Many representations show artificial intelligence has an end state or a more mature state. In fact artificial intelligence needs to be infused at every stage of the software engineering lifecycle. Machine learning and augmented intelligence manifests as the ability to infuse intelligence into existing phases, states end processes.
5. #Robotics in its virtual (read RPA, #IntelligentAutomation) and physical forms (spectrum from individual robot helpers to #EmbodiedCognition in a room or building) will become mainstream after undergoing the above four churns.
6. Data and Content #AutomatedCuration will become more practical, tools will be developed and Data Sets be more cleaned and curated and readily available.
7. #AITrainedHealthcare will be challenged by the Arsanjani Laws for Cognitive Systems and will require cross checking between AI’s trained on different IntelligenceSets and KnowledgeSets before recommendations start to take firmer hold.
8. Research and funding for using #BiasIndicatingAlgorithms will expand as pressures on #AIEthics mount and more content will be publish that will include a #confidenceFactorForNonValidatedClaims, to decrease #fakenews #alternativefacts #rumors. This will be a new #AIBattleground
9. #AI will be used increasingly for political clout in a covert and overt fashion to engage in traditional fields of engagement. #AIEthics
10. #AIEducation and #DataScienceEducation will continue exponential expansion and popularity, it will be infused into traditional software emngineering ciricula as Academic Institutions see the value of integrating BigData, Data Science, ML and AI with traditional Software Engineering curricula.
#AI #DeepLearning #ML #MachineLearning #ArtificialIntelligence #RPA #IntelligentAutomation #Chatbots #VirtualAssistants #Robotics #EmbodiedCognition #DataScience #BigData #Cognitive #CognitiveComputing
contextual computing 