On the 30th of November 2022, OpenAI released ChatGPT – a model that managed to bring AI to the forefront of every newspaper and change the outlook of the general public on AI for content generation.
Now, it is difficult to imagine a person who has not heard of Artificial Intelligence, as many disciplines have been fascinated by this topic for decades. From humanoid robots threatening to take over the world and others trying to prevent them (looking at you, Terminator), to our reality where AI is used to predict protein structures, or to help you daily in the form of Siri or Alexa. This opposing view of AI leads to confusion and its presumed dangers manage to distract from true ones.
For decades there has been a fear of AI and people were often apprehensive, but now comes along a pretty, user-friendly interface and all the inhibition seems to fade away. The thing about ChatGPT is that the technology behind it existed for a while, but the interface was a lot less pleasant. Now, however, it has become accessible – so more people are playing with it. And that’s good, don’t get me wrong – I love AI because it has the potential to open many doors, but people need to understand what it is to use it sensibly. It is not a search engine, so please don’t ask it for a weather forecast or a historical fact. It will give you some nonsense answer because its job is to compose sentences, not fact check that they are true.
Another bit of warning from yours truly is: don’t put in your full name and/or address into the prompt. Once you input the information, it will stay within the model and somebody with malicious intent can use it for bad, bad things.
We also have the issue of what the models was trained on, and the answer is The Internet (which we all know is not the safest or most considerate of places). So, the model can generate prejudiced texts, and the user must be aware of that and treat it is as a tool to help create, rather than it being the sole, unedited creator. This is also the part where I’m going to talk about ethics. As a STEM student I am more focused on “can” rather than “should”, but this way of thinking is dangerous and the ethical issues behind AI should not be disregarded.
AI has been riding a way of popularity recently, but as with everything mainstream, the popularity fades. As a society we tend to overestimate the short-term impact and vastly underestimate the long term. Everyone and their mother played with ChatGPT, they made it write about dogs and poems about biology, but what most didn’t consider is its capability to write code.
One of the most significant benefits of AI-generated code is that it speeds up the development process. In traditional software development, developers need to write code manually, which can be time-consuming and prone to errors. With AI-generated code, developers can generate code automatically, saving time and reducing the risk of errors.
Yet this is where the danger lies, because this means that scam emails will increase exponentially and become more dangerous as they get even more targeted and appear even more credible. Fear not, though, my favourite reader – you are not doomed to live a life of fighting off scams and talking to the “costumer care” of made-up banks, because the battle between good and evil has not been won yet. What society needs to do is invest in the good software engineers and companies who work to protect your cyber security, because the one thing that evil has the upper hand in is money.
In its essence, AI can help us build so many new and useful things that will improve your lives, but it is a two-sided coin. The technology exists, so don’t fear it – instead, understand it, play with it and it will no longer be a black box of mystery. In my experience of working with it for my EPQ on AI, it should be seen as a tool that supplements your work – it is not an autonomous mind that can be left unchecked.
So, have fun: ask AI to write you a sonnet about haemoglobin, a speech about ballpoint pens in the style of Shakespeare or ask DALL.E to generate you a painting of a space cat in the style of Van Gogh. Just don’t forget that GPT-4 (as of this month) isn’t a toy, but a tool capable of things we didn’t think a computer would be able to do.
If you’d like to discuss these issues further, do come along to Tech Futures Soc where digital innovation, ethics and tech philosophy are discussed every fortnight! Week B Tuesday, Innovation Room, 4pm.
