I've only dabbled in AI so far, using it to create some images for this blog (see the Canada-US topic for some examples), summarizing some long articles, and doing some web searches. Mostly, I've been using Gemini as that's easily accessible from Chrome and seems to work better than Microsoft's Copilot.
Recently a friend recommended Duck.ai, created by DuckDuckGo, home of the well-known browser of the same name. Duck.ai is a simple web-based front end to several top AI platforms and provides a high degree of privacy for its users by stripping metadata and not retaining users' information.
A menu lets you select one of several models: GPT-6 Luna, GPT 5.4 mini, Claude Haiku 5.5, Mistral Small 4, gpt-oss 12B, and Gemma 4 31B. Another menu lets you choose modes between Reasoning, which is slower and reaches usage limits faster, and Fast. A Tools menu lets you create an image, search the web, and customize your response (set tone, length, and behaviour of the AI). It's a simple, clean interface.
Last week, I was reading a post on the Eye on the Storm blog about how October was one of the warmest month's on record in the US. It included the following chart.
It's clear that temperatures are increasing, but is the trend straight line or exponential? To me, it looks exponential, but I wanted to confirm that. If I had access to the original data, I could have plugged it into Excel and figured that out, but I wasn't going to try entering 260 data points manually from the graph. (More about this later).
So I gave Duck.ai the URL of the graph and asked it to create a new graphic with the best fit curve. This is the result. It took less than 10 seconds. The curve is clearly exponential.
After thinking about it a bit, I realized that I might have biased the result by asking for a best fit curve instead of a best fit line. I asked for that instead and this is the result. Again, it's clearly exponential though not as steep.
I was curious to see if it could show me trend lines for the summer and winter data points, and yes, it could.
As you can see, the curves are diverging, which means that the spread between summer and winter temperatures is increasing and summer temperatures appear to be increasing more than winter. From what I've read about climate change, that is the case.
I used the Yale Climate Connections image as the visual reference; I did not pull the underlying data directly from NOAA. The generated charts may not reproduce the data points or fitted curves accurately. For reliable summer and winter fits, I’d need the underlying NOAA data (or a data table) and would calculate the curves from that.
Further questioning revealed that it didn't have access to the NOAA data and I would have to provide a link for it to use the NOAA data directly.
Using Google, I quickly found the website with the relevant data. Asking Duck.ai if it could use this data provided this response.
Yes. NOAA’s Climate at a Glance provides U.S. national temperature time-series data, which can be downloaded and used for separate seasonal fits. The page I found describes the data as monthly time series; I couldn’t load the exact page directly just now.To match the graphic, download the national temperature data for the same date range and send me the CSV. I’ll group the months into summer and winter, fit each series, and make the chart. I’d use June–August for summer and December–February for winter unless the graphic defines seasons differently.
That's far more work than I want to do for something like this, but it's worth knowing that it's possible and relatively straightforward. I was impressed tht it suggested how it would analyze the data if I provided it.
When I was taking math courses in university, such a task might have been given as an assignment and probably would have taken a few evenings of work. Now it takes minutes, or less.
Getting back to Duck.ai's features, I should note that it retains your sessions with the AI, which is useful. You can delete individual chats or turn chat history off.
Duck.ai does not publish what its usage limits are for free accounts and they reset each night at midnight UTC. Usage limits for paid subscribers reset weekly on Sunday and users can check their usage limits during the week. Unfortunately for me (but fortunately for my pocketbook), paid subscriptions are not available in Canada.
Of the free AI tools I've used, Duck.ai is the clear winner for both its ease of use and its privacy protection. If they offer a paid subscription in Canada, I will seriously consider signing up.






