Google's new AI Model Gemini

7 years ago, I started writing as a hobby (something other than very mind-numbing university essays). I wanted to write articles about interesting things I came across.

I remember reading an article about Google acquiring an AI company called Deepmind. The article talked about the tests Deepmind conducted on their AI which prompted me to write an article relaying my thoughts.

Google introduced their latest AI model called Gemini a few days ago. When I first saw the introduction video, I was reminded of Deepmind and, funnily enough, it was indeed the Deepmind team that worked on this new Gemini AI. At the same time as introducing this new AI model, they released videos on YouTube demonstrating its capabilities in various tasks. The most impressive demo was where the AI interacted (seemingly in real-time) with a video. That video showed the AI interacting seamlessly and smoothly with the video which to me signals the next step in what most people would consider real AI (of course, our only references for “real AI” come from pop culture).

Artificial intelligence and machine learning are used everywhere in ways that are not overtly obvious to most of us. They're used in search engines, self-driving cars and even with tools like Grammarly, which uses a version of OpenAI (the AI that powers chat GPT, among other things).

Of course, when talking about AI, we always have to talk about the potential risk — despite the fears mostly coming from movies we watch. Even though these fears come from pop culture, they are not unfounded. We all know that a powerful being ruled by logic might deem us humans as irrelevant or even a threat. For now, most “AI” systems I see seem like machine learning only. While machine learning is a subset of AI, it is not the idea of AI that we are accustomed to. Machine learning is where a computer tackles a specific task and learns and improves within that task. The AI systems we are all thinking about would be able to learn and interact like humans, meaning it is a broader application of AI.

The demo video that I mentioned earlier is one that actually makes this Gemini model seem like the AI that we are all thinking about. In this demo, the AI is interacting with video, rather than text or audio only. The interaction is smooth and natural and is close to the popular idea of AI.

I have a YouTube channel where I react to interesting things (though I have mainly been reacting to videos of one freestyle rapper so far). When I saw the video introduction of this AI I just had to go and record a video talking about it.

I watched this video before doing more research about the Gemini AI or the demo videos, so I did not realise that Google came out after releasing these demos saying that their videos were edited. Of course, when seeing things like this, we always have to be sceptical because they are not live videos. I did have some scepticism, but the excitement was a much higher priority so I decided to watch the whole video. I did notice that there was something in the video that could indicate editing (go to 22:03 in the video above) but I was still hoping that the demo at least had some truth to it because it would mean amazing improvements to AI and future technology.

I am very excited to see where this goes, but what do you think? Is this a good innovation, or are we heading towards a dystopian future where AI and technology will rule everything?