WN #16

Chatbot Wars: Part 3

Author

Kyler McClary
February 19, 2024

Chatbot Wars: Part 3

The AI chatbot wars are heating up once again. Over the past few weeks, we’ve seen major updates from some of the biggest players in the space as they try to compete with OpenAI’s ChatGPT. Both Microsoft and Google re-branded their chatbots and added a paid subscription tier, Perplexity gave its paid version a big upgrade, and OpenAI made some updates of its own to ChatGPT. All of these platforms still have free versions, but you’ll have to pay to access the best features and models.

I’ve been writing about these updates in the “Bulletin Board” section of the newsletter, but it’s worth taking a moment to pause and recap what’s new, putting it all together to help you assess which platform to use and whether or not a paid subscription might be worthwhile. Here’s an updated version of the chatbot chart, then we’ll dive in and break down each platform a bit further:

ChatGPT (OpenAI)

What’s New: Made ChatGPT Voice available to all users (free and paid); Launched GPT Store for Plus users; Added @ mention feature to pull custom GPTs into existing threads (Plus users); Launched ChatGPT Team for collaborative workspace, higher usage caps, and increased control of data; Improved ChatGPT’s memory by allowing it to learn from one chat to the next, which will make it more personalized and helpful over time (select users only for now).

Why would I use ChatGPT’s free version?: ChatGPT is great for a variety of tasks, it’s simple to use, and there are no usage limits. You can also give ChatGPT custom instructions to help better personalize its responses, and its app has full voice integration, meaning you can have a verbal conversation with it (I use this feature all the time in the car, in the kitchen, or when I just don’t feel like typing).

Why might I upgrade to ChatGPT Plus ($20/month)?: ChatGPT Plus is the Swiss army knife of AI chatbots. It uses the industry-leading AI model, GPT-4, which is a significant upgrade from the free version’s GPT-3.5, and gives you the ability to upload and generate images, search the web, upload your own files, write code, create and browse custom GPTs, and connect with external apps through plugins. Its creators, OpenAI, are at the cutting edge of the AI space and new AI capabilities tend to come to ChatGPT Plus first while other companies play catch-up. That said, if you had asked me a month ago which subscription you should put $20/month towards, ChatGPT Plus would have been the hands down winner, but recent updates from competitors have me questioning if that will be the case going forward.

Copilot (Microsoft)

What’s New: Re-branded from “Bing Chat” to “Copilot”; Launched a Copilot app; Split into Copilot (free) and Copilot Pro ($20/month)

Why would I use Copilot’s free version?: Copilot’s free version still gives you access to numerous features that require a paid subscription on ChatGPT and other platforms, including the ability to upload and generate images, along with limited access to OpenAI’s GPT-4 model (GPT-4 access during non-peak hours only). Copilot can also pull info from the live web via Bing search and cites its sources. If you use Microsoft products, you’ll also continue seeing more and more Copilot integrations. For example, Windows now includes a Copilot button in the taskbar on the desktop, and if you use the Microsoft Edge browser there is a Copilot side panel that can summarize, cite, and answer questions about the webpage or PDF you have open (see “The Lab” section below), in addition to its other features. There will even be a Copilot key added to keyboards on new laptops. Finally, if your school has an eligible Microsoft account, all faculty (and higher ed students 18+) have access to commercial data protection when logged in to Copilot with their school account, meaning all data stays in-house.

Why might I upgrade to Copilot Pro? In addition to Copilot’s free features, Copilot Pro offers faster responses, faster image generation, priority access to OpenAI’s industry-leading GPT-4 model, and integrates into Microsoft 365 apps like Word, Excel, Outlook, and PowerPoint. However, this applies only to personal and family accounts. If your school has a Microsoft 365 subscription they can add Copilot for Microsoft 365 (faculty and staff only) for $30/person per month which includes the app integrations, the ability to integrate your own data, and increased data privacy controls.

Gemini (Google)

What’s New: Re-branded from “Bard” to “Gemini”; Uses new Gemini AI model; Added image generation capabilities; Split into Gemini (free) and Gemini Advanced ($20/month); Launched a new app for Android and integrated Gemini into the Google app for iOS

Why would I use Gemini’s free version?: If you used Google’s Bard chatbot in the past and were unimpressed, join the club. But the recent re-brand to Gemini also comes with a new AI model, Gemini Pro, which is a significant upgrade. Gemini is connected to the internet through Google, and, like Copilot, offers image uploads and generation for free. Though it doesn’t cite sources by default, there is a button to use Google Search to check the accuracy of the output. If you are using a personal account, Gemini can also connect to your Google Workspace (Gmail, Drive, Docs) along with apps like YouTube (see “The Lab” section below), Maps, Flights, and Hotels. If you find yourself using Google products more than Microsoft ones, this might be the chatbot for you.

Why might I upgrade to Gemini Advanced?: Gemini Advanced (available only for personal accounts at the moment) gives you access to Google’s newest and best AI model, Gemini Ultra, which approaches the capabilities of GPT-4, and even outperforms it in some areas. The $20/month includes the benefits of Google One Premium, such as 2TB of storage, and it will soon include Gemini for Workspace, which will integrate Gemini Advanced into Gmail, Drive, Docs, and Sheets, but that feature isn’t quite ready yet. In the meantime, you can try Gemini Advanced with a two-month free trial.

What’s New: Perplexity Pro ($20/month) users can now switch between leading AI models including GPT-4, Claude 2.1, and Gemini, and can now upload and generate images

Why would I use Perplexity’s free version?: First of all, you don’t need to log in or create any sort of account to start using Perplexity, you can just jump right in. Perplexity crafts its AI responses by searching the web and including citations, making it easy to trace the information back to its source. It finds related images and videos for you as well, and you can also organize your threads into collections that you can share and invite others to collaborate on. If you want its search to be a bit more focused, you can set it to only look through specific sources like published academic papers, YouTube, Reddit, or Wolfram Alpha, or you can take off web search altogether and just let it write from its pre-trained data, similar to ChatGPT or Claude. It also allows you to upload your own text or PDF documents (up to 3 per day).

Why might I upgrade to Perplexity Pro?: Perplexity Pro adds in features like image upload and generation, as well as unlimited file uploads. Perhaps its coolest feature is that it gives you the ability to switch between AI models so you can try your prompt with GPT-4, Gemini, Claude 2.1, or other leading models to see which one best handles your task. If you want to test out different models but don’t want to sign up for a bunch of different accounts, this might be appealing to you.

Claude (Anthropic)

What’s New: Upgraded their model from Claude 2.0 to Claude 2.1

Why would I use Claude’s free version?: Unlike the other platforms mentioned, Claude’s free version gives you access to its top AI model, Claude 2.1, and outperforms ChatGPT’s free version in a number of areas. Another advantage is that it lets you upload up to 5 of your own files at a time (text, PDF, CSV), whereas ChatGPT charges you to do this. Its impressively large context window means it can read up to around 75,000 words at a time. Claude has been my go-to when I need to work with a PDF or DOC file, but the free version also has usage limits that can make it difficult to rely too heavily on it.

Why might I upgrade to Claude Pro ($20/month)?: While Claude’s free version is powerful, it does have usage limits that vary based on the length of your conversations, the size of the files you upload, and the number of other people using Claude at that time. Claude Pro increases your usage limits and gives you priority access during high traffic. You also get early access to new features and the ability to switch between different versions of Claude. Claude Pro isn’t the most comprehensive paid plan out there, but if you become a Claude fan it may prove to be worthwhile.

(Note: Of course, your interactions with AI might not include any of these chatbots, but rather a third-party, education-specific platform like MagicSchool, Diffit, or Curipod that utilizes the underlying AI models mentioned here, but packages them into an education-specific interface (for a fee, of course). I’ll be doing a write-up on these in an upcoming newsletter.)

Gen Z will be ready for the AI Revolution

AI is set to transform the world we live in, even more so for the students we teach. Here’s why one university president says graduates will be ready for it, an argument that seems to be corroborated when you talk to Gen Z students themselves.

Re-thinking Science Education

AI is already changing how we do science. How do we ensure science education reflects this? The National Science Foundation is taking steps to address this question with the launch of their EducateAI Initiative to bring AI educational experience to K-12, higher ed, and adult learners. This coincides with the recent announcement of broader AI in education initiatives, including a Congressional bill pushing for increased investment in AI literacy, and a National Institute of AI in Education at Sacramento State University.

What’s Next for Writers?

One of the big discussions surrounding the emergence of generative AI technologies is the potential impact on the writing profession. In an opinion piece for the LA Times, one writer explains why she isn’t worried.

📌 Google is Moving Fast

Between starting this newsletter and hitting send, Google already announced a huge update to their Gemini Pro model. Gemini Pro 1.5 is only available to select developers and enterprise users for now, but its main feature is an industry-leading context window (essentially how much info it can handle at once), that’s over 30x larger than the current iteration of Gemini Pro. This means that, in one go, it will be able to process up to an hour of video, 11 hours of audio, or over 700,000 words (and Google claims to have achieved 10x that amount in its research and testing).

📌 Focusing on Image Generation

The chatbots mentioned above (except Claude) all have image generators built in, but if you just want to use the image generation feature on its own you can do that as well by going to Google’s ImageFX or Microsoft’s Designer (both of which have some extra features that you don’t get in the chatbot version), and I’ll throw in an extra one: Meta’s Imagine.

📌 OpenAI’s Jaw-Dropping Text-to-Video Demo

I was going to do a little write-up on the emerging competition in the AI text-to-video space, including projects like Pika.art, RunwayML, and Google’s Lumiere, but then OpenAI came out of nowhere and blew them all out of the water with the introduction of Sora (available only to a few select users at the moment). If you are the type that is fearful over the speed at which AI is developing, you may not want to look at these examples of what Sora can do.

Using Gemini with YouTube Videos

As mentioned above, Google’s new Gemini chatbot can work with YouTube (for now you have to use your personal account, not your school one). Here’s an example of using Gemini to save time previewing YouTube videos to show students. I first asked it how a video specifically addresses the topic of segregation in South Africa:

Then, I asked it to identify some potential opportunities for discussion (not all timestamps are accurate, but the content is):

A couple of notes:

  1. When it works well this feature is nifty, but it’s still inconsistent. Sometimes it gives me what I want and sometimes it tells me it can’t help, even when using the same video link and prompt. Be persistent (and also check out the alternative drafts it produces).

  2. It will generally do better with shorter videos than longer ones, but that could be changing soon with the release of Gemini 1.5 (see “Bulletin Board” section above)

  3. Of course, you should always review the actual video yourself before using it in your classroom, but this can serve as a helpful tool in deciding which video to commit to and/or giving you specific ideas and look-for’s as you watch it.

Using Copilot in Microsoft Edge

I wrote about Microsoft’s Copilot AI above, but I want to highlight one specific feature that people often miss: using Copilot in Microsoft’s Edge browser (the default browser on Windows devices). When using Edge, you can open Copilot by clicking on the Copilot logo in the top right corner of the browser. You can then use it to interact with any webpage or PDF you have open, and it can even direct you to specific information on the page (e.g., in the example below when I ask it for key quotes from the article, not only does it list them, but when I click on a quote it highlights the corresponding text in the article).

More Ideas…

This collection provides a great peek into how AI is being used in various higher ed classrooms, along with honest and helpful reflections. Lots of good stuff in here whether you are in K-12 or higher ed.

That’s all for this week! If you appreciate the content in this newsletter, consider subscribing for free, or sharing with people in your network who may find value in it. If you are looking for more, feel free to check out the archive to catch up on any editions you may have missed.