Are you tired of AI making things up? This wizard only answers based on peer-reviewed research.

Generative AI is everywhere, whether it is used as the cornerstone of a service, to create applications, or to improve functionality within them. But with so many new products coming out all the time, which ones are worth trying?

We’re here to help you with a new regular series covering the latest and greatest AI apps and services you should know about.

One of the biggest problems facing AI assistants is the vast set of “knowledge” they are trained on. That’s why you’ll find a reference to glue being particularly tasty on pizza or non-existent legal cases. While these hallucinations have become less ridiculous over time, they are now much more sneaky and difficult to identify with a tired eye. That’s why I started including Consensus in my workflow whenever possible.

The consensus answers the question: “What if Google Scholar was an artificial intelligence assistant?” combining millions of peer-reviewed research articles to provide a broad overview of diverse topics. This makes academic content much more accessible to the average user who may not have the time or mental acuity to read articles from cover to cover.

It also does not simply state the facts, but highlights references in a separate panel carefully numbered to correspond with the details of the answer. After reading your key takeaway, I can click on a reference to get an overview of the article, view the metadata, or download the full document if available.

In particular, Consensus can be used without an account, which is very positive given that many other AI products impose this requirement, although logging in offers certain benefits. On the one hand, AI offers a much more in-depth literature review for members and various other methods to weigh data scattered across multiple articles through a Consensus Meter. If you’re looking to win a particularly heated argument based on science, here’s how you do it.

Of course, given Consensus’ focus on peer-reviewed research, it’s not exactly a direct contrast to Claude, ChatGPT or Gemini. However, that’s not really their intention. I certainly wish Consensus was available when I was in college, when Zotero was the most powerful tool on my Celeron laptop.

Other new AI applications and services you should know about!

There are many other great new AI applications and services, including:

In case you don’t know, there is a small tournament taking place this summer. Is it called World Cup or something like that? Naturally, many fans flock to stores to buy football shirts at this time of the decade, but how can you tell if that kit is authentic or fake?

I recently stumbled upon an app called KitLegit, which aims to help fans do just that. Take a photo of the shirt with your phone’s camera, upload it to the app, and let it do its job. If the jersey is authentic (or rather, if the app believes the kit is authentic), it will issue a certificate of authenticity. Given the experience of some online football fans, KitLegit is not a 100% foolproof verification method.

While you should always hold apps, especially those that leverage AI, with a healthy dose of skepticism, it’s helpful to use an app that points you in the right direction.

Not everyone enjoys the big G looking over their digital shoulder, so while NotebookLM is certainly my go-to tool and AI-powered smart laptop of choice, it may not appeal to those searching on Google or interested in hosting their own solution. That’s where Open Notebook comes into play.

It’s effectively an open source, self-deployable version of NotebookLM, giving users more control over how their data is stored, where it’s stored, and which AI models analyze it (18+ are optional).

Given these available options, Open Notebook is a much more advanced tool. You will need a place to house it and the patience to learn how to use it or the intelligence to do so.

Either way, it’s great to see an open source alternative to a tool that was previously exclusive to Google or didn’t exist at all.

I spend an inordinate amount of my free time reviewing guidebooks and atlases, planning future road trips around my country that I really hope to take. Now I enjoy manual planning, but I recently stumbled upon Mindtrip, and it may be the most elegant trip planner I’ve used yet.

It’s hard to describe Mindtrip, but it’s much more like an AI-powered Wanderlog that takes your directions and transforms them into travel skeletons than a travel-oriented ChatGPT clone. The service provides everything you need to plan a trip, including suggestions for sensible overnight stops, popular attractions along the route, and other tips and tricks.

The Explore section also provides more details about these attractions and encourages users to ask questions. For example, if you were a tourist visiting Cape Town, you could ask Mindtrip about the main routes to the summit. When planning a theoretical trip to Kruger National Park, I was asked if I would prefer to stay in park accommodation or a luxury lodge. Clean!

It could certainly use a bit of usability improvement (it often feels too clunky compared to its rivals) and I wish I could manually drag the map route to the cities I want to travel through. However, given the amount of value you can get from Mindtrip for free before setting up an account, it’s worth trying it out on your next adventure.