More and More Talkative

A couple of weeks ago I wrote about ChatGPT, OpenAI’s interactive artificial intelligence which allows a user to have eerily human conversations with an (almost) all-knowing artificial counterpart. Since then, it’s garnered significant interest both from other players in the field of AI, and the public leveraging its skills. Microsoft, of whom I suspect you have heard, even committed to a multibillion dollar investment in the tool and its parent company. ChatGPT has come out of the gate hot and has only gained speed since. The most sincere form of flattery is of course imitation, so what alternatives to this platform do we have coming down the pipe?

What I think it’s important to understand first is what makes ChatGPT so impressive. It’s easy to show a system a huge dataset, let’s say Wikipedia, and ask it to digest that data. You’d then at a very basic level have something which was very knowledgable on a wide variety of subjects. What’s difficult is having that system, in this case an AI, answer questions using that knowledge in a succinct and relevant way. In a nutshell, this is where artificial intelligence is different from simply searching for keywords. An AI is able to understand context. That’s not to say it understands a topic like an actual, real human does, but it is capable of delivering answers which may fool you into thinking it might.

Unsurprisingly, one of the companies rumored to be looking to launch a competitor is Google. They’ve been working with AI for quite some time, but have yet to release something as public-friendly as OpenAI’s new tool. It’s likely, given Googling something is the preferred choice for around 85 percent of all web searches, to cause quite a stir and offer better understanding of current events than ChatGPT, which is limited in how recent the data it’s using is. One can see the appeal of, when Googling a question, being offered a natural sounding explanation alongside your usual web results. Being Google does have such a high profile, they’re reportedly being exceedingly cautious about the accuracy of their tool prior to release. After all, any AI is only as useful as the data you “teach” it with. Google of course isn’t the only one with AI research under its belt. Meta, the parent company to Facebook, has had a dedicated team working on artificial intelligence for years now. Their Sphere product, which is a powerful tool which uses AI to prevent the spread of misinformation on social media, could easily harness its knowledge for answering queries in an unbiased manner. Microsoft meanwhile have chosen the “join them” approach when building their AI-powered toolset. Since becoming major investors in OpenAI, they’ve integrated the power of ChatGPT into their business collaboration platform, Teams. You can now, albeit with a premium subscription, have Teams generate highlights, take notes and create tasks from any meeting you hold using the platform. A summarized overview of a meeting is so much more useful than trawling through a transcript; this is genuine and easily identifiable real world benefit.

2023 is set to be an exciting year for AI-enabled programs and seeing how they can make us more efficient. Fewer repetitive tasks, less time spent trawling for research purposes, easier checking for mistakes. I for one am fully engaged with how we might leverage the accessibility of this technology to make our day to day lives easier. While it’s fun to ask a machine existential questions and observe its reply, the more valuable question is how we might put such a machine to work.