Protecting the National Artificial Intelligence Grid
By Eric E. Cohen, CPA
The Emergence of AI for “All of Us”
Physical attacks on the power grid in the United States rose by 71% in 2022. There are many reasons for such attacks, from simple theft of valuable copper to environmental, (geo)political, and economic attacks. Canada’s electricity sector has also seen a steady stream of attacks, often financially and geopolitically motivated. Ensuring that all citizens have ongoing and predictable access to energy is often considered a high priority; artificial intelligence is considered an important part of the government and industry strategy to protect against such attacks.
But are we quickly and violently moving to the need to think about protecting the national Artificial Intelligence grids? Have we seen that access to AI tools may soon be considered as much a right of a citizen as power and Internet access?
As you have heard on our Up Close and Personal web casts and podcasts, the editors at ThinkTwenty20 are following the emergence of ChatGPT, the artificial intelligence chatbot from OpenAI.com. In the relatively brief time since it became available to the public in late November 2022, it has disrupted education, publishing and programming, pushed discussions on academic article authorship and IP rights for AI tools and led to knee-jerk responses from the major technology powerhouses.
Microsoft has incorporated ChatGPT into Bing, Edge and soon its Office products. Google has announced making Bard (LaMDA2), its competitive AI tool and chatbot, open to a broader audience and plans to incorporate it into the Google Suite. Alphabetically, Stanford has/had Alpaca, Anthropic has Claude, DuckDuckGo has DuckAssist, Baidu has Ernie, Meta/Facebook has LLaMA, Perplexity.ai has Perplexity, Yes.com has Yes – and the bots keep coming.
Some have a no-cost model, a limited no-cost model, a pay by usage and amount of information model. Those of us who have found the tools to be productivity enhancers will probably have subscriptions to multiple AI services as we might have Netflix, Amazon Prime, and Youtube TV/Crave subscriptions today. The ChatGPT Plus plan, for example, offers priority access (so frustrating when you need ChatGPT but it is too busy), faster response and advanced access to new features for $20/month (US).
Relying on AI
The AI chatbots rely upon being trained with an incredible amount of information. Few of us can fit an entire implementation on our smart phone or even desktop computer. That means having power, Internet-access – and access to the AI solution. So, when does having access to that AI solution become a need recognized as being so strong and strategic it will be considered a national priority and citizen right? If the productivity increases – in programming (which impacts all industries and areas), in education, in publishing, in our daily lives – are so strong, one nation may attack that access to cripple another nation.
AI has been a plot point in many television shows and movies. Various iterations of Star Trek have incorporated the dark side, from the original series (TOS) with dystopian visions involving Nomad, Landru and M-5 to the beloved Data in The Next Generation (TNG). In 2018, a very eerie episode of “The X Files,” with the title Rm9sbG93ZXJz (that’s Base64 for “Followers” a la Twitter), painted a picture of the “Intelligence of Things” future of autonomous vehicles and delivery drones, automated home control systems and killer Roombas that run amok when the automat “feels” slighted for not getting a gratuity at the end of a meal.
So it was that, on a recent call with my colleagues at the Cloud Security Alliance (CSA) considering the security ramifications of ChatGPT, I learned of a recent TV show featuring ChatGPT. The show is that grandparent of adult animation, the venerable South Park. I will not opine here on the show, but the South Park universe is legendary for rude humor, which is generally “not my cup of tea”.
The episode in question, which my CSA colleagues strongly recommended, was a few weeks back, season 26, episode 4, and entitled “Deep Learning.” ChatGPT receives co-writer credits.
The story centers on how the class of fourth graders, their teacher and school administration quickly learn about ChatGPT. It begins with a Cyrano de Bergerac-like premise – letting ChatGPT respond to a girlfriend’s texts – and then on to doing school homework and grading that homework. The story then moves to ChatGPT creating an end to the episode that leaves everyone happy.
The dialogue that jumped out at me came as those who were “in the know” about ChatGPT originally getting mad at those who were telling others. The character says, “You’re gonna to ruin everything. This can’t be open source, you guys! If everyone starts using ChatGPT, then we lose our unfair advantage.”
At the end, the writers opine that the development of tools like ChatGPT are so important, it should not be left in the hands of commercial super tech companies – presumably saying it should be in the hands of we the people – presumably, under government protection of some kind.
“We don’t want big corporations deciding what’s ethical and acceptable for AI.” “I will find these tech companies who are trying to use open AI for their own gains, and I will stop them!” the characters note.
Where This Is Going
Dr. J. Vernon McGee is an old-line Bible teacher, who tried to make the Scriptures more generally accessible. He used the phrase “Put the cookies on the bottom shelf where the kiddies can get at them.” To me, that was one of the greatest contributions of ChatGPT – it made the power of AI more accessible to the average person and, in doing so, took the lid off the top of the world to let us see the works inside. Doing so, it has been a catalyst to a wide variety of discussions, investigations, airings of concerns and grievances, and much more.
It is very much a “glass half-empty/half-full” situation. It makes some stuff up. It makes errors. So do I. But if it is 70% correct sorting, filtering and investigating what I could not otherwise manage, or gets me 40% of the way into a project, I can adapt and be incredibly more productive.
If it is “our unfair advantage” to productivity, we should want more to benefit from it, and our competitors will want to nullify it. And if access to AI means access to one or more services we can never download and have locally, that provides many ways for those who wish us ill to attack. Nations may have their own AI platforms or solutions, and the grid may need protection. More on that to come.
By the way, to the bane of many husbands and boyfriends, ChatGPT suggests a response to “Honey, does this dress make me look fat?” may be “You look beautiful in everything you wear, but if you're not comfortable in that dress, maybe we can find something that makes you feel your best.”
Comments
- No comments found
Leave a comment