This piece is my own opinion, drawn from over 20 years of experience working in AI.
In 2001 I was in the final year of my A-levels and deciding which degree path I wanted to follow. The boom of the internet had really piqued my interest in technology, and I knew I wanted to follow some sort of technology-related path. My dad took me to open days at several universities, but one stood out because of the robotics demonstration they were showcasing. In that moment I fell in love with robotics and AI and decided this was where I wanted to study. So, I applied and was accepted.
I started in 2002 and back then, while AI had been around for about 50 years, it was not well known. This was down to a lack of access to datasets to train the models, as well as insufficient hardware capacity. There was no stance on whether AI itself was good or bad. It was more that we had to remain conscious and vigilant of the ethics surrounding it. We even had a module on the ethics of AI. Bias and AI discrimination were at the forefront of the learning. We also learnt about Alan Turing’s story and his fate in spite of all he had done for the field. Discrimination in AI has always been there whether societal or algorithmic.
As companies like Google started to emerge, their motto was “don’t be evil”, and even OpenAI started as a not-for-profit looking to produce AI for good, something Karen Hao speaks to in her book Empire of AI. These principles were rooted in the Computer Science community and its commitment to open source. However, when capitalism comes into the picture this quickly changes.
When we were studying, we always looked at AI as a way of helping humans solve problems. Some examples were robots in disaster zones for search and rescue operations, where it was too dangerous for humans, space exploration, neural networks for medical technologies and so forth. I thought, wow, we can do great things here. There was talk of robots that could behave like humans and more human-like replication, but it was deemed far-fetched, as AI could only ever be a crude approximation of human intelligence. For me, the takeaway was that it could be used to solve specific problems in useful ways. However, with the good, there was always the bad. I remember my robotics lecturer telling us he had been approached to design military robots for war zones but had declined because of his ethical principles. As a naïve 20-year-old, I was really shocked by this. For me, the ‘destruction of humanity’ angle has always been there.
When a tech bro tells us AI is going to destroy humanity, what does this mean? I was stunned when I read the headline. It has been known that there is potential for this and, yes, we should name it and think of it as a possibility, even if it is a small percentage. But why is this white man, who has made millions from AI and people like him, who have acted as gatekeepers of AI, saying it now? I was stumped. I mean, if they truly believe that they do not have to build it, right? They have been talking about Artificial General Intelligence for a long time. No one wants that. I do not want that. What is the angle?
I had been trying to figure out the angle since this made the news. I mean, it is already causing destruction, is it not? The advancement of military robots, the impact on the environment including water usage, the displacement of communities to build data centres and mine for hardware and the loss of critical thinking now generative AI is so publicly accessible. I could not figure it out. But the wheels started turning when I listened to the latest episode of The World, the Universe and Us by New Scientist last Friday. They described it as criti-hype, a term coined by Lee Vinsel, where criticism exaggerates a technology’s power and dangers in a way that ends up feeding the same hype created by its boosters and distracting from problems such as bias and high energy use. If it is such a threat, they have the power to switch it off. It was also described as a dead cat strategy: throwing something shocking on the table so everyone talks about that instead of the problems you would rather they ignored.
I wondered about the dead cat strategy, but I was still struggling to figure it out. I thought, who better to ask than one of my closest friends from university, who studied the same course as me: Dan. We had a lengthy discussion and he shared his thoughts. He described AI as insatiable, with an endless hunger for resources, data and power. His take was that if you do not give it more, someone else will. He gave the example of the iPhone. The original R&D costs were very high, but the pay-off was that once it was in production, those costs would drop and profit margins would increase. This isn’t possible with AI because to get better, the models need more data, more computing power and bigger data centres, and each new model has to be bigger than the last to keep up with competitors, which means it never ends. The tech bros are realising they cannot make a profit, and data centres are costing more and more. By ringing the alarm, the belief is that they are trying to cover up their own failings and spark panic so governments will invest and start to regulate without the tech bros having to do any heavy lifting. It started to make sense, and once our conversation ended this very thing started appearing in my social media feed. One post from NYC-DSA Tech Action put it simply: AI alarmism and AI hype are the same story, both designed to keep our attention on a science-fiction future and away from the harms happening now. I know people will say the phones are not literally listening, but the point is that the signals are being picked up somehow.
Then this week Sam Altman, the head of OpenAI, stood on a stage in San Francisco and told a room full of business people that the world is right to be afraid of AI. In the same breath he said we should trust companies like his to do the right thing and that the industry can regulate itself. Mark Zuckerberg and Jensen Huang from Nvidia backed him up, with Huang saying they don’t need new laws at all. They tell us that we should be scared of what they’re building, but leave them in charge of it? Not everyone in the industry agrees, and Anthropic’s boss is calling for governments to step in, but either way the conversation is still all about their imagined future. The water usage, addressing bias and the communities being displaced for data centres didn’t get a mention. The line that really stuck with me came from Dario Amodei, the head of Anthropic. He said one of the biggest surprises of the AI boom was how fast these companies grew and how quickly they became central to everything. That’s the part I’d be worried about, that a handful of companies becoming so embedded in our lives, our work and our governments that switching it off gets harder every day. Right now, they still could. They’re choosing not to.
Why am I writing about this when all of this is already on social media? Mostly because I am frustrated. Since I embarked on this journey in 2002, the white male tech bro voice has always been centred, and it continues to be centred now. People like me have never been given funding to develop ethical AI (we end up self funding), a principle I strongly stand for. Now, because of this alarmist narrative, my work will be tarred with the same brush the tech bros are painting with, and it is even less likely I will receive support, financial or otherwise. I also don’t believe in totalism that it’s either all bad and will destroy us or completely amazing and will revolutionise the world.
My plea is for people to hold their own minds, to critique where it needs critiquing, and not to fall into the panic narrative driven by tech bros for their own agenda. I continue to question and check my own role in the development of AI and I’m open to changing my stance should evidence emerge that all of it is bad. If this happens, I will discontinue my work, however currently I am doing meaningful work in this space as are others such as Christian Ortiz who built Justice AI and Shae O. Omonijo who built the Critical Thinking Bot. My company is using AI to help detect lung cancer, where catching it earlier can save lives. In hospitals, avatars can give staff a safe way to build their skills before they are in front of real patients. In psychology and therapy, it can help trainees understand experiences such as racism or neurodivergence, so they can learn and practise without asking real people to relive painful moments for the sake of someone else’s education. This is what we set out to do all those years ago at university, using AI to help humans solve problems, but once again those of us who are marginalised are left out of the conversations. Most of us don’t need AI to write for us or to summarise our emails and phone message notifications, but even that has it’s uses for people who have accessibility needs.
I’ve decided to persevere and to keep fighting for ethical and meaningful AI. Everything has the potential for good and bad. It all depends on what we choose to do with what is at our disposal, tech or otherwise.