Eric Horvitz ✓

@erichorvitz · Seattle WA USA · joined 12 Mar 2011

Chief Scientific Officer, Microsoft

31 543Followers

1 596Following

3 001Posts total

11KViews on collected posts

Against accounts of the same size

9 posts from the last 90 days, next to the 10K–100K follower range. below its peers on both reach and engagement.

Median views775this account1 456median for 10K–100K

Reach, %2.46%this account4.50%median for 10K–100K

Engagement, %0.48%this account1.88%median for 10K–100K

Metric	This account	Median for 10K–100K	Ratio
Median views per post	775	1 456	0.53×
Reach (views ÷ followers)	2.46%	4.50%	0.55×
Engagement rate	0.48%	1.88%	0.26×

Others in this range → Compare with another account → How these benchmarks are built →

Growth & engagement

How the posts we collected actually performed: views and reaction rate post by post, what the audience did with them, and where the follower count goes.

Views per post

1.1K19 Aug

3.3K20 Aug

2.9K4 Sep

945

676

690

775

525

Last 9 collected posts, oldest on the left. The scale is logarithmic: one post can outrun the rest a hundred times over.

Engagement rate per post

0.09%19 Aug

0.48%20 Aug

0.00%

1.24%4 Sep

0.95%

1.04%

0.43%

0.90%

0.19%

Reactions — likes, reposts, replies and quotes — divided by views. Median for 10K–100K accounts is 1.88%.

What the audience does

Likes71.6%63 in total

Reposts10.2%9 in total

Replies4.5%4 in total

Quotes4.5%4 in total

Bookmarks9.1%8 in total

Share of every reaction we collected for this account. Replies mean argument, reposts mean endorsement, bookmarks mean the post was worth keeping.

The follower curve appears once this account has two daily snapshots — we take one a day, and this one is on its first.

Latest posts

Calling out concerns with trends on understanding and guiding AI. 525 views · 1 likes · 0 reposts · 0 replies 04 Sep 2026 Tandem training: An approach aimed at incentivizing models to be understandable. Considered "understanding" by weaker models--but had human interpretability of machine intelligence in mind. 775 views · 5 likes · 2 reposts · 0 replies 04 Sep 2026 Astra's model card shows that monitoring for abuse via CoT may become increasingly difficult. Beyond efforts on alignment & refusals, impt. to invest in maintaining monitorability. R&D on efforts like "tandem training" will be important. More from @cervisiarius below. 690 views · 3 likes · 0 reposts · 0 replies 04 Sep 2026 6/ This is why @erichorvitz and I warned in @ScienceMagazine about "A narrowing window to understand AI". Human understanding—even if only partial and high-level—is crucial, and it won’t emerge on its own; it must be an explicit training objective. https://t.co/LIWOBJLt5o 676 views · 5 likes · 0 reposts · 1 replies 04 Sep 2026 7/ One proposal: tandem training (@ashton_anderson @erichorvitz), where a strong "senior" (Astra?) co-generates solutions with a frozen weaker "junior" (~humans). To get reward, the senior's reasoning must be something the junior can pick up and continue. https://t.co/DXqZeaTUkb 945 views · 7 likes · 0 reposts · 1 replies 04 Sep 2026 1/ The state-of-the-art way of monitoring state-of-the-art models is to read tea leaves. Left: a model reasoning to itself. Right: the same model talking to other agents. Every frontier lab depends on this, @sama @DarioAmodei @demishassabis https://t.co/umj7vrFeWQ 2.9K views · 29 likes · 4 reposts · 1 replies 04 Sep 2026 @erichorvitz Thanks for sharing. The research was clearly ahead of its time, but great material for product innovators today. 44 views · 0 likes · 0 reposts · 0 replies 20 Aug 2026 Here's a demo of Lumiere, the research prototype that inspired Clippy (the Office Assistant): https://t.co/oMMRB4AYZb As background: The shipped version operated quite differently. The Office team decided to execute the Bayesian inference component only when there was a 3.3K views · 13 likes · 2 reposts · 1 replies 20 Aug 2026 Nature Digital podcast on clinician-AI collaboration. https://t.co/aco4MvqJvE 1.1K views · 0 likes · 1 reposts · 0 replies 19 Aug 2026