Tyler Benster
@tbenst
Followers
3K
Following
370
Media
171
Statuses
1K
Stealth startup founder. Neuroscience PhD from Stanford: silent speech interfaces. GP of Asimov Ventures: 3D printing, robotics, deep tech.
Stanford, CA
Joined June 2009
Our silent speech preprint is live! Using a cross-modal training technique enhanced by LLMs, we set a new state-of-the-art for silent speech (12.2% word error rate, open vocabulary) and brain-to-text (8.9% WER; Rank 1 on Brain-to-Text Benchmark '24)
4
19
96
Generate videos in just a few seconds. Try Grok Imagine, free for a limited time.
393
663
3K
Bullish AF for neurotech. Capital rushing in faster than talent though. Neurotech wins when rigor is rewarded as much as storytelling—otherwise it’ll be a lost decade of hyping modalities that don’t cut it.
FT Exclusive: OpenAI and its co-founder are preparing to back a company that will compete with Musk's Neuralink by connecting human brains with computers.
0
0
2
I think the main miss with GPT-5 was actually timing. o3 felt like the “GPT-5” level jump in intelligence and was probably the right moment for the PR. GPT-5 vs 4o was a huge change as was o3 vs 4o, but GPT-5 and o3 feel very similar.
0
0
6
Key moat for the proprietary models is multimodal. "For those seeking multimodal support, built-in tools, and seamless integration with our platform, models available through our API platform remain the best option.". here's to hoping someone releases adv voice mode.
0
0
0
OpenAI's first open-source LLM since GPT-2 is out! gpt-oss impresses as expected in benchmarks. The hidden gem here is the release of the Harmony Response Format + code for rendering system prompts and conversations. Unsexy but super valuable for devs
gpt-oss is out!. we made an open model that performs at the level of o4-mini and runs on a high-end laptop (WTF!!). (and a smaller one that runs on a phone). super proud of the team; big triumph of technology.
1
0
3
As a long-time user of em dashes, it saddens me that everyone now thinks that AI writes for me when in fact I've been an LLM since age 5.
0
0
2
Spoiler: What’s wrong with Alexa speakers, Siri on phone, Humane AI Pin, and Rabbit R1 isn’t a lack of sun protection. It’s that talking out loud to your device is cringe and embarrassing.
6
1
32
RT @BasedBeffJezos: As soon as Neuralink offers subvocalization decoding it will immediately be worth getting for everyone. Imagine subvoc….
0
31
0
Soon, the only need for human involvement will be constantly telling AGI to accept cookie policies.
0
0
1
If o3 tries to answer a question by searching and citing something you wrote on the internet, you just got cooked.
0
0
0
I love that Brain-to-text '25 now includes both vocalized and silent speech!
kaggle.com
Decode intracortical neural activity during attempted speech into words
1
0
5
Lol maybe Humanity’s Last Exam should be called The Last Time You Play Bar Trivia. > What was the rarest noble gas on Earth as as a percentage of all terrestrial matter in 2002? Answer: Oganesson.
0
0
1
Benchmarks are amazing! But curious how well it works in the wild. I tried a few queries and interactions with search is still far behind OpenAI/Google/Anthropic, harder to tell for pure reasoning given how 99% of my use of reasoning models now also uses search.
🚀 We’re excited to introduce Qwen3-235B-A22B-Thinking-2507 — our most advanced reasoning model yet!. Over the past 3 months, we’ve significantly scaled and enhanced the thinking capability of Qwen3, achieving:.✅ Improved performance in logical reasoning, math, science & coding
0
1
2
RT @aran_nayebi: 🚀 New Open-Source Release! PyTorchTNN 🚀. A PyTorch package for building biologically-plausible temporal neural networks (T….
0
38
0