By Xiling “Celine” Gu
“Real AI doesn’t exist,” said Bruce Epstein, a senior software engineer and knowledge transfer expert from AEBIS Corp. He sat down for the interview with Disruptive Technologists, and shared his view on what constitutes real autonomous AI. Bruce believes the prevailing misconception about current AI programs is that they are intelligent but that “AI products may look like they are intelligent, but they are just acting upon a predetermined script.” Intelligence however, comes with autonomous decisions made by instinctive judgement based on massive tacit knowledges, implicit and explicit information.
To Bruce Epstein, major AI digital assistance products on the market are transactional conversational robots that can convert limited human vocabulary into command, and act upon predetermined algorithms. During the process, it’s the users who learn how to talk to the machine, instead of the machine learning how to take orders. Bruce gives an example of how a digital assisting program should work to achieve the level of real AI, “I should be able to ask a calendar program, ‘when will be a good time to visit my mother’? The program should be able to infer what the user is really asking, and autonomously generate algorithms to fulfill this command.
In Aebis, engineers are transferring knowledge that’s “naturally” embedded in human brains, the knowledge that had become instinct, intuition or automatic reflex. Aebis’ expertise lies in transforming such tacit knowledge into the form that can either be used to transfer to another person, or to present in a language that can be programed into a smart algorithm for the AI environment. Aebis’s efforts aim to empower AI technology so that it will eventually function similar to the powerful human brain’s instinctive decision-making mechanism.