Technologies worth trusting

AI has arguably made some significant changes in our modern society, but one thing that's talked about a lot within the AI industry is that there are a lot of risks associated with AI. For example, AI scientists and people in the AI industry regularly talk about their p(doom), and top scientists worry about it. Yoshua Bengio, who is the most cited across all industries, worries about AI risks and regularly talks about AI safety. He even signed an open letter to ask AI labs to pause AI training for six months [1]. Even the CEO of Google's DeepMind, Demis Hassabis, said that "It's going to be 10 times bigger than the Industrial Revolution, and maybe 10 times faster" when he was talking about how fast AI will change society [2]. While some people worry about the risks of AI, others ignore the good it has done. One example is cancer screening. AI has made cancer screening a lot better. In a study performed by the MASAI Trial research team, the researchers found that AI-assisted mammography screenings help detect more cancers. The AI cancer screenings found 338 real people with cancer, whereas the human-only cancer screenings only found 262 people with cancer. This is a 29% increase in detecting cancers [3], so while AI has helped a lot in different industries like the medical industry, it would also be unwise to forget some of the major accomplishments in AI safety. OpenAI and Apollo Research worked together and tested a technique called deliberative alignment. This was done to reduce AI from scheming. The amount of scheming was able to be reduced from 13% to 0.4% in the ChatGPT o3 model [4]. These types of positive leaps in AI safety are extremely good, but are these leaps in AI safety enough? Well, in my professional opinion, I say MAYBE because with technology, there will always be an associated level of risk. Some technologies have bigger risks than others, but managing those risks carefully and delicately is how we can minimize the risk of extreme levels of failure or bad outcomes. Technologies can be light-sensitive regulation, and more important decisions at AI companies should be decided by the people, which is how we properly minimize risk