Google's vision for the Pixel 9 is great, but it doesn't escape my biggest issue with AI

I haven’t found myself using generative AI regularly yet, even with Gemini dominating all of Google’s products lately. With the Pixel 9 series, AI is more important than ever, but I find myself frustrated by the same fundamental flaw in generative AI that even the Pixel 9 can’t escape.

This issue of 9to5Google Weekender is part of the rebooted 9to5Google newsletter that highlights the biggest Google stories with added commentary and other details. Register here To get it delivered to your inbox early!

From the beginning, and especially in light of the use of generative AI to “improve” search, my problem with generative AI has always been that the technology is often inconclusively correct. You might get an answer right 10 times, but in my eyes it’s not worth it if you get one answer later that spits out incorrect information as absolute truth.

It’s a problem that generative AI, at least Google’s, hasn’t really been able to solve yet.

Misinformation seems to be the obstacle that most affects Google’s AI ambitions. You can’t think about this topic without thinking of “pizza glue.”

On the Pixel 9 series, Google is putting AI front and center, and even after just a week of use, I’ve already come across multiple instances of this confident misinformation that remind me why I rarely use AI tools.

One example of this this week was an afternoon meeting. The restaurant I was supposed to meet at was closed at the time we scheduled the meeting (around 3 p.m.), so on the way in, I asked about other options close to that first location that were already open. Specifically, the question was, “Can you show me other restaurants that are open near you?” [insert restaurant name]”?”

Instead, Gemini literally spat out the site I just mentioned.

When I corrected the AI and asked what options were available, it repeated the same place. Then I repeated that the listed place is closed, so what happened? It said the same thing again.

Meanwhile, opening Google Maps and searching for 20 seconds gave me the answer I needed with much less frustration.

The primary goal of generative AI in an assistant is supposed to be to get better results that can be derived, not just point them out. But Gemini continues to fail on this front all too often. In this example, it’s not that Gemini doesn’t have the information. In fact, Gemini even directly mentions the location’s hours. It just fails to say how long the location takes. Used That information.

*Gemini had all the information he needed to give me at 2:45 p.m., but he completely ignored it.*

I found the same to be true in Pixel Screenshots.

One thing I was excited about with this app was the ability to take screenshots of online orders and ask the AI to find estimated arrival dates, or find the date you placed an order. But it’s pretty bad at it. For example, I asked it to find the date I bought a shirt — I placed an order on August 15 and took a screenshot of the checkout page — but instead it pulled up a completely irrelevant Amazon return and claimed I ordered the shirt on September 9, weeks later. Further attempts after I added more screenshots to the app didn’t get any better. At one point, it said I ordered the shirt on August 19, the day I asked the question, while the screenshot below shows the app saying I placed the order on August 9, days before I started using the phone and a date that isn’t mentioned in the any From the screenshots I took.

But it’s easy to overlook this. Pixel Screenshots only uses the information in the screenshots. It can’t take advantage of tracking numbers, and it can’t read my email, so the information it has is limited. Ultimately, I don’t mind that. If it can’t, I understand. That’s not easy at all!

*I still can’t figure out where this date came from.*

The problem is that it confidently gives an answer that may be correct, or may be greatly misleading.

This is why I simply don’t see the appeal of things like Gemini Live. Google’s conversational AI assistant is certainly impressive in the way it speaks and reacts to what you say. But that doesn’t stop it from spewing out wildly incorrect information that it presents as absolute truth. In my hands-on experience with the Pixel 9 this week, Andrew Romero had an issue with Gemini Live where the AI insisted he was planning a trip to Florida despite him repeatedly saying that it wouldn’t happen. This is clearly an AI hallucination, but it really reinforces my distrust of these products.

Until these issues are resolved, I think there’s a huge gap in Google’s vision for AI in the Pixel 9 series that can’t be solved with small compromises. We can argue about practicality and whether anyone wants these features, but the fact is they’re happening. I just want to be able to have some confidence in them.

Top News of the Week

Pixel 9 Reviews

Following its announcement earlier this month, here are our first reviews of the Google Pixel 9 series. Stay tuned for a camera-focused look at the Pixel 9 Pro XL, a Pixel 9 Pro Fold review, and reviews of the Pixel Watch 3 and Buds Pro 2 coming soon.