GPT vs Bard, and Google’s Keynote AI Announcements

why Bard may be better than GPT, and Google’s AI announcements from its May 10 2023 keynote

Google’s keynote was big for AI, as it introduced its PaLM 2 AI language model — the tech behind Bard.

Some key points from Google’s keynote:

📧 “help me write” feature in Gmail (now available for waitlist) will write a full email from a simple prompt

👨‍💻 Bard can support coding capabilities in 100+ languages and can chat directly with other Google apps and third-party extensions (Chrome extensions — this is a big deal)

🖼️ Bard announced its partnership with Adobe and its ability to create images — big for the design industry

If you’d like to watch a summary, here’s Engadget’s 18-min review.

GPT vs Bard?

Bard is quickly becoming a strong opponent to GPT — it’s essentially showing that it can train stronger and faster than GPT — not to mention it’s connected to the internet in REAL TIME (GPT is limited to 2021 and prior)

🕰️ Bard operates in real-time, while GPT is limited to data from September 2021 in prior. This also means Bard can provide more specific and accurate responses to anything related to local businesses and services (note: I do suspect that GPT will come to be real-time within the next 6 months )

💰 Bard is completely free to use — GPT is freemium model

🕵️‍♂️ Bard’s got better privacy — Bard uses “federated learning,” which allows it to train on data without ever actually seeing the data. This makes it a better option for companies that require high levels of security (perhaps Samsung allows its employees to use Bard given its GPT ban…?)

💬 Bard’s got better text quality — it understands and produces more complex sentence structures, as well as nuanced and metaphorical language — essentially, it’s more sophisticated and agile in its speaking (which begs the question — can it bypass the current AI that’s working to determine if AI or a real person makes something?)

🗣️ Bard can communicate and translate in 3 languages — GPT is English only

💪 Bard is more efficient (uses less resources) — the better text quality above is a result of Bard being built on “Reformer” neural network technology; this tech is also the reason it’s able to create responses with less computational resources (and perhaps not use as much water as GPT…?) — and faster.

📝 Bard can summarize text from webpages directly — GPT cannot.

Like this summary? Follow me on LinkedIn for more updates.