Google announced PaLM 2 at I/O last week, along with updates to Bard and integration of AI across Workspace (Docs, Sheets, Gmail). This isn't Google catching up to OpenAI. It's Google making a serious play for enterprise.
Here's what matters for your firm.
PaLM 2: Better Than GPT-4 in Some Ways
Google's new model is comparable to GPT-4 on most benchmarks. Sometimes better, sometimes slightly behind. The key differences:
- Coding: PaLM 2 is genuinely better at code generation and debugging
- Math: Better at mathematical reasoning
- Logic: Better at complex reasoning tasks
- Multilingual: Supports more languages and does better on multilingual tasks
For professional services, the coding advantage matters less. But the math and logic improvements could be useful for analysis-heavy work.
Bard Updates
Bard now uses PaLM 2 instead of the older LaMDA. It's noticeably better. More accurate, better at reasoning, more reliable.
Google is rolling it out wider. More countries, more languages, more integrations.
The competitive message is clear: "GPT-4 isn't the only good model. Bard is now competitive."
Google Workspace Integration: The Real Play
This is the significant announcement for your firm.
Google is integrating AI directly into:
- Docs: "Help me draft," "improve this writing," "summarize this document"
- Sheets: "Analyze this data," "create a chart," "find patterns"
- Gmail: "Suggest a response," "draft an email," "summarize this thread"
- Meet: "Summarize this meeting," "generate action items"
This is exactly what Microsoft is doing with Office. And it's coming soon.
What This Means for Your Firm
If you use Google Workspace: AI features will be built into your tools soon. You'll need to think about data governance (same conversation as with Microsoft).
If you use Microsoft: You're still ahead, as Microsoft has been pushing this harder. But Google is catching up fast.
For both: You're going to have AI in your core productivity tools whether you formally approved it or not. Plan for that.
The Competitive Space Now
We're officially in the era of mature AI competition:
- OpenAI: GPT-4 via ChatGPT and API. Fast iteration, strong developer community
- Google: PaLM 2 via Bard and Workspace. Integrated into enterprise tools, massive distribution
- Anthropic: Claude. Safety-focused alternative, growing enterprise interest
- Others: Mistral, LLaMA, and open-source models gaining traction
None of these are going away. The market is supporting multiple winners.
What You Should Do
If you're on Workspace: Start thinking about the inevitable AI integration. What's your data governance stance? What will you allow? What will you prohibit?
If you're on Microsoft: Similar conversation, but you have more time since the rollout is slower.
For both: Test Bard. Compare it to ChatGPT and GPT-4 on your real work. See if it's useful for your firm. Understand the differences.
Document your policy: Whatever tool your team uses, document what's approved and what's not. Update as things change.
The Opportunity
The fact that you have choices now is good. It means the market is maturing. It means better tools are coming.
Don't wait for one clear winner to emerge. The winners are already emerging, and you need to be using them now to stay competitive.