Abstract
The world seems to have become addicted to the new digital drug of AI. Although AI has been around for a long time, it has only recently hit the streets in its generally palatable form thanks to the likes of ChatGPT, MidJourney and DALL-E.
Stanford Professor John McCarthy first used the term Artificial Intelligence (AI) in 1955, defining the meaning as ‘the science and engineering of making intelligent machines’. We have, more recently, developed ways to program such machines to behave cleverly. In 1997, 1 the AI environment that we had developed was intelligent because it represented space in a smart way, meaning that the calculation of distance in a complex geometric environment was smart. With light and sound, the tracing of initial paths and reflections in complex spatial geometries becomes computer-intensive. Although contemporary computer power can often handle the processing of such calculations in a reasonable time, the technique of spatial data processing using tesseral methods certainly offered an efficient, intelligent alternative to conventional geometric processing. Digital computers are still better at numbers than they are at shape.
The focus today is the ability of machines to both interpret and learn, at least somewhat like human beings do. What seems to be forgotten, or perhaps taken for granted, is that much of the contemporary AI, and certainly the populist AI, is enabled by the Internet and the information repositories around the world that it connects. That is where the fuel for the ‘intelligence’ comes from. It was only in 1989 that Tim Berners-Lee stitched together the technologies that gave us the Internet. DALL-E and its older cousin GPT-3 rely on being able to scrape the data that stoke up the intelligent engine from Internet repositories.
What supervised learning and reinforcement learning technologies will need is additional forms of intelligence, and that is difficult to achieve. What data is subject to copyright, and perhaps more of a challenge, what data is good and relevant data? I asked ChatGPT to write a short essay about Peter Rice and his work at the Sydney Opera House. It did pretty well with 80% but couldn’t resist telling me that Sydney Opera House is a reinforced column and beam structure. What makes Sydney exceptional is that it isn’t conventional in that way, but how does an artificially intelligent system learn to know that when the phrase ‘reinforced concrete’ is so frequently linked with the idea of a rectilinear frame?
Outside our research area, Universities, more generally, are much vexed by whether AI is something that they should be battling with or embracing. One university in the UK has recently lost a legal case, having been proven to have falsely accused a student of using AI to write an assignment. 2 The AI detection wrongly concluded that one of its relatives wrote the assignment in question. Ironically, the student was in a Law School.
It also seems that AI systems are not currently smart enough to know that they are plagiarising or infringing copyright in an unacceptable way. 3 It will possibly take a while for AI systems to develop a legal mind and hence to know when they are borrowing with implied consent or paraphrasing sufficiently and when they are stealing without payment.
Given all of this current attention to artificial intelligence, it is inevitable that all the sections of papers in this issue employ AI in one form or another. The ten papers in this issue are divided into four sections. In the section entitled Computer Mediated Design, Zhou and Park’s AI-augmented Sketching of Conceptual Design is the most explicit in calling on the aid of Artificial Intelligence as we currently know it. But as the reader will find, McCarthy’s definition of artificial intelligence pervades all of the research articles here present in this issue. What also pervades our computationally enabled lives is the idea of cryptography, and Richard Coyne’s book on this subject, reviewed in this issue, is both an entertaining and salutary read to supplement reflections on machine intelligence and security.