Sign in

Chao Wang

@excel-wang.bsky.social
25 followers 20 following 87 posts

Associate Professor in health and social care statistics at Kingston University. PhD in econometrics.

PostsRepliesMedia
Chao Wang @excel-wang.bsky.social · 29/09/2026
If you think LLM is the way to achieve general intelligence (AGI?), think again. Using chess/Go as an example, LLMs are still far behind top human players, let alone top chess/Go AIs such as AlphaGo (which is not an LLM). x.com/i/grok/share...
x.com
LLMs Poor at Chess and Go
General-purpose LLMs are mediocre at chess (roughly beginner to intermediate club level at best) and very weak at Go. They cannot beat professional human players or top specialized AIs like Stockfish,...
100
Chao Wang @excel-wang.bsky.social · 23/09/2026
Study design does not necessarily dictate the level of evidence, as often depicted in "evidence pyramid". Context is important. See the example below (youtu.be/3pxudE0GNAo?...). Many people thought Study 2, a cross-sectional study, is the weakest. However Study 2 is arguably the strongest.
110
Chao Wang @excel-wang.bsky.social · 14/09/2026
Interesting article. It shows LLM cannot discover causal relationships purely based on the data without experts’ input.
0112
Reposted by Chao Wang
Jess Calarco @jessicacalarco.com · 31/07/2026
For example: When I teach qualitative methods, I ask my students which of these foods are sandwiches, and there's always lots of disagreement. So, we talk about how they can explain their criteria and try to persuade others to agree with them. But they can't objectively prove they're right.
Six images of food items: tacos, wrap, hot dog, banh mi, avocado toast, burger.
11566
Reposted by Chao Wang
Adam Kucharski @adamjkucharski.bsky.social · 05/08/2026
Two fictional (presumably AI) stories in recent X posts, now repeated as a Google summary. No need to tell tall tales about your achievements when the AI ecosystem will now do it for you…
65119
Chao Wang @excel-wang.bsky.social · 07/07/2026
You will get a better chance with closed-source feature-rich (i.e. not heavily relying on third-party packages to do anything useful) software.
010
Reposted by Chao Wang
Chao Wang @excel-wang.bsky.social · 21/05/2026
Tried a few models and both Gemini 3.5 Flash (Extended, screenshot1) and M365 Copilot GPT 5.5 (Think Deeper, s2) got the correct answer. Some reported Claude Opus 4.7 is also fine. Yes more advanced mode may not always be better, but they tend to be more correct than instant/quick/... mode.
001
Chao Wang @excel-wang.bsky.social · 11/05/2026
Nice benchmark "Which LLM writes the best Stata code?" www.khaledeltokhy.com/benchmarks/
000
Reposted by Chao Wang
Sarah O'Connor @sarahoconnorft.ft.com · 26/03/2026
In spite of all the talk of Claude Code and Codex meaning the end of humans writing code, software job adverts are actually going up, according to @jburnmurdoch.ft.com's crunching of millions of job ads for this week's The AI Shift www.ft.com/content/7325...
20565174
Chao Wang @excel-wang.bsky.social · 26/02/2026
Use LLMs to automate statistical reports drwangstatsconsulting.wordpress.com/2026/02/26/u...
drwangstatsconsulting.wordpress.com
Use LLMs to automate statistical reports
Large Language Models (LLMs) can enhance productivity by automating code generation for statistical reports. Using Gemini and Stata, users can input analysis code, which Gemini converts to Stata co…
000
Reposted by Chao Wang
Kevin Mitchell @wiringthebrain.bsky.social · 21/02/2026
The strongest version of this illusion I’ve seen! Absolute head-wrecker!
25388122
Reposted by Chao Wang
Jason Concepcion @netw3rk.bsky.social · 06/12/2025
Couldn’t Sam Altman just ask ChatGPT how to make itself profitable
684610762
Reposted by Chao Wang
Thomas House @tah-sci.com · 24/11/2025
This by @whippletom.bsky.social is brilliant. A little dose of epistemic humility goes a long way. www.thetimes.com/comment/colu...
thetimes.com
We’ll need good data next time or lockdown arguments multiply
At the start of the pandemic two professors, Martin Landray and Peter Horby, did something that should have been banal but was also rare: they tested drugs
074
Reposted by Chao Wang
Darren Dahly @statsepi.bsky.social · 24/10/2025
We tried to tell y'all to stop calling everything "AI" many years ago and you just wouldn't listen and now the poor machine learners must also suffer alongside the statisticians 😜
68911
Chao Wang @excel-wang.bsky.social · 16/11/2025
I asked ChatGPT how to calculate confidence intervals without researching and sourcing third-party R packages. Here’s the response.
000
Reposted by Chao Wang
Paolo Crosetto @paolocrosetto.bsky.social · 12/11/2025
What is the most profitable industry in the world, this side of the law? Not oil, not IT, not pharma. It's *scientific publishing*. We call this the Drain of Scientific Publishing. Paper: arxiv.org/abs/2511.04820 Background: doi.org/10.1162/qss_... Thread @markhanson.fediscience.org.ap.brid.gy 👇
8334238
Chao Wang @excel-wang.bsky.social · 16/10/2025
Using Excel to bootstrap a simple linear regression and visualise the process. drwangstatsconsulting.wordpress.com/2025/10/16/u...
drwangstatsconsulting.wordpress.com
Using Excel to bootstrap a simple linear regression and visualise the process.
Excel is widely used and one of the most familiar professional software among students regardless of their background. Excel can help students, in particular those without statistics background, to…
000
Reposted by Chao Wang
Compton Scatterbrained @comptonscatter.bsky.social · 01/10/2025
The future is bleak: AI "researchers" will send questionaires to AI "respondents" to collect data on what they "think" about Topic, poorly summarise the results and generate a paper for a mill. AI peers will review and leave comments. Other AI will cite the paper (and nonexistent ones). Wooooo.
1131
Chao Wang @excel-wang.bsky.social · 17/09/2025
New meta-analysis of effects of 20mph speed limit in the UK shows 𝙣𝙤 statistically significant impact on killed or serious injured (KSI) crashes or injuries on sign-only schemes. t.co/WSvmnbLJ1n
000
Reposted by Chao Wang
Darren Dahly @statsepi.bsky.social · 21/07/2025
I think it was a target trial emulation of difference in differences for a propensity score matched analysis with Bayesian borrowing of historical controls of digital twins based on synthetic data from the real world.
3197
Reposted by Chao Wang
Peter Tennant @pwgtennant.bsky.social · 17/07/2025
Once you truly understand causal inference theory, it's very hard to win an honest applied grant because: * Many questions can't be answered * Almost everything that sounds big and exciting is snake oil * Most reviewers won't understand * you're competing with the bandwagon-jumping cynics
4697
Chao Wang @excel-wang.bsky.social · 17/06/2025
More on “Simple method works equally well…” and do you really need SuperLearner? drwangstatsconsulting.wordpress.com/2025/06/17/m...
drwangstatsconsulting.wordpress.com
More on “Simple method works equally well…” and do you really need SuperLearner?
I recently came across some interesting tutorials on TMLE. This reminds me of my previous post on comparing different estimators of treatment effects (see here). The default setting in the tmle R p…
000
Reposted by Chao Wang
henrybemis13.bsky.social @henrybemis13.bsky.social · 03/05/2025
Half the data science influencers i followed in 2019 are now LLM/AI influencers, which leads me to believe late 2010s data science scene was all hype
032
Chao Wang @excel-wang.bsky.social · 23/05/2025
New blog post ““Zombie statistics” and strivers” drwangstatsconsulting.wordpress.com/2025/05/19/z...
drwangstatsconsulting.wordpress.com
“Zombie statistics” and strivers
There is a recent article on identifying potential students in Royal Statistical Society’s magazine Significance ( which I found very interesting. A response letter ( is also worth reading. T…
011
Reposted by Chao Wang
Khoa @khoavuumn.bsky.social · 03/05/2025
7 habits that make you look like a professor.
51395
Reposted by Chao Wang
Maarten van Smeden @maartenvsmeden.bsky.social · 26/03/2025
If the goal is to describe or to explain, true high colinearity between important covariates may be a fact of life and difficult to overcome. Quick-fix solutions may result in bad confounding corrections. Bottom-line: quick fix solutions like PCA are often a poor man's solutions to colinearity
093
Reposted by Chao Wang
Jon Mellon @jonmellon.bsky.social · 03/12/2024
I get the impression that polisci is now pretty sold on LPM over logit. What's the go to citation making the case for this?
7477