2h ago

Google hides detailed thinking traces on Gemini behind a three-dot menu, yielding minimal summaries that omit key model process details such as web search verification

Wharton professor Ethan Mollick flagged the reduced visibility into model steps.

β€”β€”0β€”β€”
Original post

Google has hidden thinking traces on the Gemini site. You have to use the 3 dot menu to pull up summaries, which are so minimal as to be unusable. Did it do web searches? Did it check results? You can't tell. This makes Gemini unsuitable for any serious work you need correct.

4:02 PM Β· May 19, 2026 View on X

(I originally wrote that Gemini removed thinking traces. People in the comments pointed out that they are accessible from a menu, so I did a new tweet. But I cannot believe how useless the thinking traces are. It makes Gemini outputs completely unauditable and thus untrustworthy)

Ethan MollickEthan Mollick@emollick

Google has hidden thinking traces on the Gemini site. You have to use the 3 dot menu to pull up summaries, which are so minimal as to be unusable. Did it do web searches? Did it check results? You can't tell. This makes Gemini unsuitable for any serious work you need correct.

11:02 PM Β· May 19, 2026 Β· 12.8K Views
11:04 PM Β· May 19, 2026 Β· 3.9K Views

I find this continually frustrating. Gemini 3.5 Flash is excellent, as is Gemini 3.1 Pro. But you absolutely cannot use them for any serious purpose right now, especially for any enterprise work. Compare to Claude or ChatGPT: you can understand what the model did & how to correct

Ethan MollickEthan Mollick@emollick

(I originally wrote that Gemini removed thinking traces. People in the comments pointed out that they are accessible from a menu, so I did a new tweet. But I cannot believe how useless the thinking traces are. It makes Gemini outputs completely unauditable and thus untrustworthy)

11:04 PM Β· May 19, 2026 Β· 3.9K Views
11:09 PM Β· May 19, 2026 Β· 4.3K Views

The crazy thing is that Antigravity does this quite well! So it isn't like Google is hiding these thinking traces because of distillation or because they are full of insane mutterings.

I guess you need to either do the work in Antigravity or not at all?

Ethan MollickEthan Mollick@emollick

I find this continually frustrating. Gemini 3.5 Flash is excellent, as is Gemini 3.1 Pro. But you absolutely cannot use them for any serious purpose right now, especially for any enterprise work. Compare to Claude or ChatGPT: you can understand what the model did & how to correct

11:09 PM Β· May 19, 2026 Β· 4.3K Views
11:14 PM Β· May 19, 2026 Β· 2.6K Views

Fascinatingly Antigravity is actually the best tool so far at providing this sort of transparency, doing something by default that Codex and Code do not: offering a summary of exactly what it did at the end of a task.

Just add this to Gemini! (But also cite sources more)

Ethan MollickEthan Mollick@emollick

The crazy thing is that Antigravity does this quite well! So it isn't like Google is hiding these thinking traces because of distillation or because they are full of insane mutterings. I guess you need to either do the work in Antigravity or not at all?

11:14 PM Β· May 19, 2026 Β· 2.6K Views
11:25 PM Β· May 19, 2026 Β· 4K Views

And the Great Powers have become aware of this thread.

Josh WoodwardJosh Woodward@joshwoodward

@iruletheworldmo @emollick Oh, let us check. Thanks for flagging!

11:24 PM Β· May 19, 2026 Β· 4.4K Views
11:36 PM Β· May 19, 2026 Β· 4.1K Views

@emollick @joshwoodward huge paper cut to fix

Ethan MollickEthan Mollick@emollick

Google has hidden thinking traces on the Gemini site. You have to use the 3 dot menu to pull up summaries, which are so minimal as to be unusable. Did it do web searches? Did it check results? You can't tell. This makes Gemini unsuitable for any serious work you need correct.

11:02 PM Β· May 19, 2026 Β· 12.8K Views
11:19 PM Β· May 19, 2026 Β· 593 Views