Since Dalle2 entered the public sphere there has been an increasing interest in regulating Artificial Intelligence. Scholars and policymakers have suggested that government regulation of Artificial Intelligence is necessary to safeguard the public interest. This is a curious and sudden change in technology industry narrative. Facebook, Google, Apple, Amazon, Microsoft, and Netflix have argued for years that their use of Artificial Intelligence to process your data is not a threat and does not need government regulation. Wasn’t it just a few years ago that Microsoft, Amazon, and Google were battling over the right to provide facial recognition services to the military? Internet Search (like Bing and YouTube), Streaming Media (Spotify and Netflix) Recommendations, Shopping Recommendations (Amazon, Shopify), Airline Fares (United, American), Digital Assistants (Alexa, Siri), Facial Recognition, Self-Driving Cars, Image-Tuning Smartphone Photos (iOS and Android), and even Smart Devices (Ring Doorbell and Nest Thermostat) all use Artificial Intelligence. So, what would it even mean to talk about asking the government to regulate Artificial Intelligence?
Generative Search vs. Artificial Intelligence
Artificial Intelligence is a general branch of Computer (or Cognitive) Science that attempts to understand the foundations of (human) perception and reasoning. Generative Search is a much more limited area within Artificial Intelligence that deals specifically with creating human-like output in the form of text, images, or media (music, code, movies, etc) based on input in the form of “prompts”. Prompts are compelling because they are expressed in common language form – you can “tell” a machine what you want and it will accurately create it. This capability has elevated Artificial Intelligence from the realm of arcane practitioners to the common user’s experience. Anyone can now use Artificial Intelligence in a meaningful and compelling way. This is what has made Generative Search the sudden wake-up call that years of data privacy warnings and academic work has failed to produce.
Turing Test and Artificial Intelligence
Textbooks of various types – History, Science, Literature – have long held that the Turing Test would be the definitive evaluation of a machine’s ability to think. The common expression of the Turing Test is that once a machine can approximate human discourse to the degree human’s cannot tell it is a machine, then it has passed the Turing Test and demonstrates de facto intelligence. Turing anticipated many of the potential arguments against his test and they are summarized nicely by the Stanford Encyclopedia of Philosophy. For our purposes, it is worth while to quote his second anticipated objection to his test, the “Head in the Sand” argument:
If there were thinking machines, then various consequences would follow. First, we would lose the best reasons that we have for thinking that we are superior to everything else in the universe (since our cherished “reason” would no longer be something that we alone possess). Second, the possibility that we might be “supplanted” by machines would become a genuine worry: if there were thinking machines, then very likely there would be machines that could think much better than we can. Third, the possibility that we might be “dominated” by machines would also become a genuine worry: if there were thinking machines, who’s to say that they would not take over the universe, and either enslave or exterminate us?
As it stands, what we have here is not an argument against the claim that machines can think; rather, we have the expression of various fears about what might follow if there were thinking machines. Someone who took these worries seriously—and who was persuaded that it is indeed possible for us to construct thinking machines—might well think that we have here reasons for giving up on the project of attempting to construct thinking machines. However, it would be a major task—which we do not intend to pursue here—to determine whether there really are any good reasons for taking these worries seriously.Stanford Encyclopedia of Philosophy, The Turing Test by Alan Turing
Generative Artificial Intelligence
In case you didn’t read the quote above, the summary is as follows: “if people really believed machine intelligence was a threat then they would stop developing it.” As of this writing, GPT-3 can already pass multiple exams designed for Humans. These are written exams, blindly submitted to professors in various areas of study, and graded as if submitted by Human students. This is an approximation of the Turing Test if anything is but keep in mind this is version 3 of ChatGPT. The current published version of ChatGPT is 4. Version 5 of ChatGPT is rumored to be ready for release before the end of 2023. Much of this is a moot point because Generative AI already either achieved “Intelligence” or fooled some very accomplished practitioners when Google created LamBDA in 2022. The fact that the world’s foremost practitioners were forced to come forward and inform us all we were wrong and LamBDA wasn’t sentient kind of misses the point. It wasn’t long before we were right back in the same place with competing claims about whether ChatGPT had achieved “intelligence”. Here it seems ChatGPT passed the Turing Test. Here it is claimed ChatGPT didn’t pass the Turing Test. Regardless, it seems we can agree it won’t be long before Generative Search does achieve “intelligence.”
Current debate all focuses on how Generative Search AI might be used to displace human authority figures. In the era of Fake News and defamation lawsuits against news organizations, it is hard to believe “misleading information” is the real concern. Stop and think for a minute: are we going to make it illegal to duplicate human capabilities by a machine? Of course not! Is the Military going to stop building drones and super space-planes that fly themselves? No way! Will we create laws that tell Google and Microsoft they can’t use Generative Artificial Intelligence in their software? That seems hard to believe as well. So, what does it mean to regulate Artificial Intelligence in the area of Generative Search? If Generative Search is really just a tool to steal Intellectual Property then perhaps the real goal of government regulation is to limit competition and existing legal liability. Recent articles in both the Economist and the Atlantic magazines suggest limiting competition as the real motivation for the recent cries for government regulation of Artificial Intelligence. Regulation would define the circumstances under which is it is not copyright infringement to repurpose existing copyrighted content for use in Generative Search. Right now, there are multiple lawsuits alleging copyright infringement against the major Generative Search Artificial Intelligence purveyors. On the other hand, most commercial Generative Search platforms have strict filters on images they can generate to limit liability from users generated media abridging copyright. Below are some images that can be generated using Open Source Generative AI (that you can’t generate on commercial platforms) and it makes it clear how Generative Search relies on existing Intellectual Property.
Don’t Restrict Generative Artificial Intelligence
Radio Spectrum licensing during the 20th Century seems a likely candidate for how Government Regulation might be implemented. There is a long and detailed history available regarding the application of government regulation related to radio and broadcast, and in particular, Amateur “Ham” radio licensing. To easily see how and why this might be a relevant analog for Artificial Intelligence, just take a look at the history of Pirate Radio. Given the complexity of Artificial Intelligence, and the many facets of life to which it already applies, treating it as “spectrum of frequency” is an apt metaphor. Artificial Intelligence is likely to be divided, licensed, and controlled by government and corporate entities for the sake of established economic interests.
Before Government Regulation for Artificial Intelligence is implemented, the public needs to make their voices heard. Generative Search is an area of Artificial Intelligence where the public stands to benefit the most because it is an area in which they can participate. Unlike Military applications, surveillance economy products, or self-driving vehicles, everyday people can participate in the creative boom that will follow Generative Search. The best way to ensure that Generative Search creates more jobs than it destroys is to ensure that anyone can use it. Corporations, early investors, and venture capitalists will assert they deserve special consideration but let’s remind them their lack of circumspection created the need for regulation. Special Consideration will be that they continue to be allowed to participate on equal footing with the people who supplied the foundational input – that is, you and me. If it wasn’t for the stored emails, accumulated art, Wikipedia entries, shared code, open source software, and decades of indexed websites, there would be no Large Language Models or Generative Search. All citizens should have equal access to Generative Search Artificial Intelligence without having to pay fees to corporate entities. Conversely, corporate entities should have to pay copyright holders for the raw input of intellectual property that fuels Generative Search. Imagine if small business owners could generate their own websites, business process automation, and even business-specific applications. What if anyone could create a compelling illustrated narrative for storytelling, advertising, or entertainment? What would it be like if Fan Fictions could become media entertainment that rivaled the (rather diminished streaming era) studio productions? How exciting if any aspiring DJ could use Generative AI to transform aspiration into inspired new music? A world where Generative Search Artificial Intelligence is set free is a world of massive and unparalleled creativity by and for the people.