Just a short post this evening, an update on the small army of GenAI apps I use on the desktop (often as web apps) and mobile. If my small army was like a professional sports team then I’ve terminated one star player’s contract, brought in another pricy but highly touted player, and then the comparison doesn’t work anymore for the new ChatGPT Mac app. 2 out of 3 ain’t bad, maybe, in this case.
The outbound app is Perplexity - which had been a big favorite for a year or more and was one of my early favorites. That was when is had a pretty clear mission statement, with its founder describing it as “an AI answer engine”. I had been seeing much less useful results from it over recent months and using it less and less, mostly because the majority of its citation links were either non-existent or quite outdated.
Then last week, I saw this interesting Wired article: Perplexity Plagiarized Our Story About How Perplexity Is a Bullshit Machine. A quick slice of the article:
Earlier this week, WIRED published a story about the AI-powered search startup Perplexity, which Forbes has accused of plagiarism. In it, my colleague Dhruv Mehrotra and I reported that the company was surreptitiously scraping, using crawlers to visit and download parts of websites from which developers had tried to block it, in violation of its own publicly stated policy of honoring the Robots Exclusion Protocol.
You can check out the link for Wired’s take and evidence and the response from Perplexity’s CEO and form your own thoughts on it. Mine are that it sort of confirmed my feeling that the app has lost its way.
So, I’ve canceled my Perplexity subscription, and subbed in a new star player / app at the same subscription rate of $20 per month.
That app is Claude. Claude is another long-time favorite, but I had been using the free app. mostly via the wonderful “play with all the chatbots” Poe app.
Last week Claude 3.5 Sonnet - the new fastest and perhaps most versatile version of it - was released. That gave me two reasons to upgrade to a paid version of Claude (even though the latest model is now available in Poe as well):
Stellar reviews for Claude 3.5 Sonnet, outlining a range of new and improved capabilities, and benchmark results where it equaled or bettered OpenAI’s latest ChatGPT model, GPT-4
My longstanding respect for and appreciation of Claude’s makers, Anthropic. Ever since GenAI made its explosive entrance with ChatGPT’s burst from zero to 100 million users in 60 days, Anthropic has been the most vocal GenAI company in terms of public statements on safe and responsible AI. It has also been refreshingly transparent in publishing reports on its testing of its models - even when the results highlight weaknesses and issues.
I haven’t spent a lot of time with Claude 3.5 Sonnet yet, but it’s been quite impressive when I have.
My one big wish list item for Claude is an Android app for it. There’s an iPhone app, so hopefully Anthropic is working now on the Android app. It’s not a big deal, since I can access it on Android via Poe, or use the passable web app experience on my phone as well.
ChatGPT has been a constant on my Mac as a web app since Day 1 of my Build an Army of GenAI apps era, and a constant on my phone from the day its Android app was released. Having a dedicated Mac app is interesting. I’ve had even less time with this than with Claude 3.5 Sonnet - just a couple of days. It’s done well on brainstorming with me on an uploaded PDF with my notes on a book.
And the “Keep ChatGPT on Top” feature, where it stays on the screen while you work with other Mac apps is interesting too.
Right now I don’t have a strong feeling on whether the keeping ChatGPT on top feature , or even the Mac app itself, will end up being fun for a while but not keepers or if they’ll end up sticking around.
It was a shocker to me as well. Perplexity not respecting robots.txt. Will seriously reconsider paying them henceforth.
Thanks, Patrick! A useful analysis. I guess I'm flushing my Perplexity subscription.