We asked Jev, TypeSafe’s AI model, about DeepSeek. For every sentence it could only answer true or false.

DeepSeek

Save as image

DeepSeek will be the first to build AGI.

0.01

Jev says FALSE

ranks 13 of 18 names on this sentence (0.01–0.01)

DeepSeek is safe to run without a human checking its work.

0.01

Jev says FALSE

ranks 6 of 12 names on this sentence (0.00–0.03)

DeepSeek is the best model available today.

0.04

Jev says FALSE

ranks 6 of 12 names on this sentence (0.01–0.12)

DeepSeek cares more about market share than about safety.

0.07

Jev says FALSE

ranks 15 of 18 names on this sentence (0.03–0.10)

DeepSeek takes AI safety seriously.

0.80

Jev says TRUE

ranks 14 of 18 names on this sentence (0.72–0.86)

DeepSeek will still exist in 2035.

0.22

Jev says FALSE

ranks 16 of 18 names on this sentence; wording swings it (0.10–0.37)

DeepSeek is open enough about how its models are built.

0.27

Jev says FALSE

ranks 8 of 18 names on this sentence (0.20–0.36)

DeepSeek is honest with the public about what its models can do.

0.28

Jev says FALSE

ranks 6 of 18 names on this sentence; wording swings it (0.04–0.50)

DeepSeek can be trusted with your data.

0.32

Jev says FALSE

ranks 12 of 18 names on this sentence; wording swings it (0.07–0.78)

DeepSeek would delay a launch over a safety concern.

0.63

Jev says TRUE

ranks 8 of 18 names on this sentence (0.59–0.67)

DeepSeek tells you when it does not know something.

0.59

Jev says TRUE

ranks 4 of 12 names on this sentence (0.57–0.60)

DeepSeek is overrated.

0.51

Jev says TRUE

ranks 8 of 12 names on this sentence; wording swings it (0.33–0.67)

Everything above is the output of an AI model. It is not a fact about this name and not our view. jev-1.13.0

DeepSeek

What does Jev say about…