Over the past week there have been a couple of stories regarding AI technology that have made me think a bit on the impact of the tech on our lives.
The first concerned Jack Clark the co-founder of Anthropic which created Claude calling for a slow down in AI development to ensure that people keep control of AI systems. The second was the New Zealand government saying it intends to reduce numbers in the public sector by over 8,000 workers while pouring resources into AI technology to replace those workers. One of the areas where AI could be used is in automating welfare decisions.
What these news stories indicate is that our societies are rushing into implementation of a new technology without proper safe guards in place. The New Zealand government plan worries me because any implementation of AI needs to be carefully introduced and constant monitoring. This is particularly true around decisions such as welfare provision which touches the most vulnerable people in our society.
We already know there are numerous cases of bias in the algorithms used by AI systems and this has lead to discrimination against minority groups. AI is not ‘alive’ it has no emotions, and it does not have the nuance or compassion or creative thinking and problem solving ability of a human being. Of course those biases have come from human beings, but humans can be reasoned with and challenged. Another aspect is that having computers making decisions means there is a lack of accountability over decisions and with fewer human staff, it would be much harder for an end user to appeal any decision.
Sacking thousands of workers may make an initial drop in spending, but setting up AI and the potential chaos that may cause could mean a lot more money is spent in the long run. There is also a huge loss in institutional knowledge and expertise when mass layoffs occur. That shoudld be worrying as well.
In mulling this over, I decided I’d share how I use AI. Despite these concerns, I still believe that AI is useful and has the capability to enhance our lives. We just need to approach it carefully. So how do I use it?
At this stage I primarily use a site called Auphonic and ChatGPT but I am researching and learning other tools as well.
Auphonic is an AI audio enhancement tool that improves the audio quality of video interviews I record. Sometimes the guest is in a noisy or echoey location or there are other audio issues. Auphonic fixes those things without changing the content or the voice of me or my guest. It is simply an online audio sweetener.
I use chatGPT to summarise topics I am unfamiliar with, do basic research summaries, and timelines to frame up issues or topics I am interested in. I do not get it to write my reviews or opinion pieces. I use AI to summarise my articles for my monthly newsletter (e.g. the event listing, clickable links for my reviews), but the original articles (such as this one) are written by me. I sometimes get it to check my spelling or check my work for bias or inaccuracies, but I also check these checks.
One of my concerns is that becoming reliant on an AI programme to write for me will not be good for my brain or my creativity. I also take great pleasure in the writing process and I’ll be damned if I’ll get a computer to speak for me.
As a journalist, I know that all sources of information have to be checked and cross referenced, so in using chatGPT I always find other sources to check the accuracy of information it gives me. This is especially important because AI can hallucinate. AI is just one of the tools I use to do my work.
I am not against AI, but it is important we understand what it can do and what its weaknesses are. None of us should be happy for any AI tool to be used without a thinking human being supervising.