The TLDR

Highlight: We’re testing out Bard’s latest updates to see if the new integrations help it close the gap.

Musing of the Week: We consider how AI could change our workforces and hiring priorities as we look to build standout teams.

💨 Is Bard Catching Up To ChatGPT?

As avid users of Google's suite of products, we were excited to hear about the new updates to Bard announced last week. The prospect of an LLM seamlessly integrated with Google Workspace seemed like it could be a game-changer, turbocharging our efficiency across emails, documents, spreadsheets, and more.

To give Bard the best chance of success, we decided to focus our testing on the use cases highlighted in the recent update announcement:

• Searching for specific information in Gmail: We put Bard through a few iterations of looking for things like receipts, travel details, or documents in our email backlog. Bard did manage to locate some of the requested receipts and documents. However, it fell flat in extracting specific, valuable details from the emails. Instead of feeling like we were interacting with a sophisticated LLM that could intuitively assist with our requests, it felt more like a slightly enhanced search function.

• Finding Flights/Hotels: Travel planning has been a popular application for LLMs, as evidenced by the many plugins released for ChatGPT. Despite Bard's direct access to Google’s travel search systems, this rendition left much to be desired, feeling like a minimal conversational layer atop the established Google Flights and Hotels platforms.

  What was most frustrating was Bard's inability to retain specifics like travel dates and destinations, even when clearly provided. Bard often ignored these details, returning seemingly random responses that actually felt more frustrating than just using the apps separately.

• Accessing Google Drive Documents: This was the tooling that had us most excited. With Bard's promise of Google Drive integration, we envisioned a streamlined process where we could casually ask about details from a doc or spreadsheet and get precise answers in return.

  In our tests, Bard was generally able to locate documents with a bit of nudging. But Bard has a fleeting memory, often forgetting the document in question or, more bafflingly, referencing an entirely different file. Instead of a smooth back-and-forth, we found ourselves repeatedly clarifying which document we were discussing, breaking the natural flow of the conversation.

Across all of the implementations, the biggest problem remains the overall lower quality of Bard's outputs. Even when it's all working properly, it just wasn't competing against the quality we've come to expect from ChatGPT. If we were comparing Bard to GPT-3 only, it would be closer since GPT-3 lacks recent data and doesn't offer integrations. But against GPT-4? There's a clear winner. Plugins more than makeup for the lack of recent data and native extensions, and the overall higher quality of GPT-4 outputs will keep us using Chat for the foreseeable future.

🧠 Musing of the Week

Hard skills, once the gold standard for most roles, are generally easier for AI to replicate. As models get stronger and more diverse, it will be difficult for humans to keep pace. If this process continues, it may force us to reevaluate our priorities when it comes to hiring and building teams to find differentiation in new places. Empathy, collaboration, and radical creativity are innately human attributes that are much harder to quantify and, thus, also harder for AI to replicate. They're the glue that binds teams, the catalyst for groundbreaking ideas, and the touch of connection that builds loyal customer bases.

Before long we could soon be scouting for people who excel in human and creative traits rather than just expertise in a particular software, technology, or technique. This runs exactly counter to the keyword-focused methodologies currently used by most AI-assisted applicant tracking systems. It seems likely that something is going to need to shift soon, and a balance between understanding how to leverage this new technology and how to build exceptionally strong relationships might just be the next most sought-after trait.

🙌 If you’re hyped about the generative AI industry specifically, here are some of the coolest roles we’ve seen this week:

• Senior Product Director, Responsible AI @

• Salesforce Creative Director, AI @ Meta

• Product Manager, AI @ GE Health

• UI/UX Designer @ Elloe AI

• Senior Manager, Customer Success @ Digital AI

🔨 Check out these other AI tools we’ve been looking at this week:

• text.cortex - Product description generator for Shopify sites

• TalkVisor - AI customer service chatbot for Shopify sites

• Cardinal - AI-powered backlog enriching your features with customer feedback and revenue data so you can best choose what to build next.

That’s all for this week. See you next Tuesday!

Lorel & Reily