ChatGPT made waves upon its release in late 2022, quickly becoming the go-to AI tool for both work and entertainment. Since then however, the landscape has evolved, and other providers have upped their game, with a range of new models on the market that are challenging ChatGPT. 

 

Known as Google's answer to ChatGPT, Bard had a somewhat shaky start, making errors at it’s launch, however, as discussed, discounting too early would be a mistake. Notably, Bard draws from more up-to-date information compared to ChatGPT-3.5. While it may still exhibit occasional accuracy issues, this is par for the course with all large language models (LLMs), so no output should be taken as gospel.

 

So, what sets Bard apart from ChatGPT? One standout feature is that a coding solution can be shared via hyperlink (something you can do for any Bard output), while you can export to Colab or Replit if that’s what you’re using. Unlike ChatGPT 3.5, Google's generative AI tool can also analyse and identify subjects within images (ChatGPT 4, the paid version can also do this). Additionally, Bard boasts a new speech function, akin to Google's translate tool, which will have various useful applications.

 

With these features in mind, Google's chatbot shines in collaborative work and we expect it’ll get better in terms of accuracy and relevance.
 

As an alternative to ChatGPT, Anthropic’s Claude has been on the radar for some time, and the recent launch of Claude 2 is powerful development.

 

Claude's standout feature is its ability to extract insights from uploaded files. For instance, we asked it to tell us what the number one hiring challenge for cybersecurity recruiters was based on a Hays report and it was able to find the correct information.

 

However it had mixed success with CSV files. It struggled to total the values in a column when the table was not at the top of a page but was able to identify figures in a simpler table, as well as total up the columns and rows. In any case, it regularly prompts the user to offer feedback, which can easily be done when provided the wrong answer.

 

Overall, Claude saves time in summarising documents and scanning reports. It offers the same functions that ChatGPT-3.5 does in terms of idea generation, although some of the information it found online was outdated, so that’s something to be wary of and is not unique to Claude – all content generated by these AI models should be further fact checked or verified.
 

How best to describe Pi? Let’s ask Pi itself.

 

“I can provide information on a wide range of topics, from history and science to pop culture and current events. I can help with decision making, by providing pros and cons and alternative perspectives. I can offer emotional support by being a good listener and offering a non-judgmental ear. And I can provide recommendations and suggestions, based on the context of our conversation.”

 

Whether you choose to think of Pi as a personal assistant or a workplace buddy its warm and informal tone makes it a welcomed companion when prioritising tasks or considering new perspectives. Pi actively listens, making it a preferred choice for those who want to share information.

 

You can discuss a stumbling block you’ve hit when writing for example, and it will gather the information needed to write a couple of paragraphs itself. We wouldn’t turn to Pi to write content, but it has a unique approach.

 

Pi may not suit everyone, but it has a wide range of uses and sets a high standard for future virtual assistants.

 

If you’re looking for a free alternative to ChatGPT, we recommend exploring each of these models. Each has its unique strengths, so it’s about finding what works best for you depending on your requirements and the task at hand. Remember that these alternatives are still evolving and perfection can’t be expected just yet. Nevertheless, they're sure to assist you in your day-to-day work.