Dave Davies

34.5K posts

Dave Davies banner
Dave Davies

Dave Davies

@OnlineInference

The Twitter profile of Dave Davies, covering SEO/GEO and AI. I spend more time on LinkedIn: https://t.co/V60S06LRcI

Victoria, BC, Canada Katılım Haziran 2008
283 Takip Edilen4.2K Takipçiler
Dave Davies
Dave Davies@OnlineInference·
@nickel88888888 @Reuters Let's see... forestry management hasn't changed in Canada, other than having more equipment and greater access to remote areas. So if there are more fires, what could it be? I mean, some studies from big oil have ruled out climate change, so it couldn't be that - right?
English
0
0
0
20
Nickel
Nickel@nickel88888888·
@Reuters Add extra tariffs for the wildfire smoke. They choke us out every year now and it's on purpose.
English
1
0
3
229
Reuters
Reuters@Reuters·
President Trump said his latest tariffs targeting Canada were not in response to wildfires, but were a separate response over trade issues, while Canadian Prime Minister Carney said he and Trump agreed to intensify trade negotiations reut.rs/4gPqUc7
English
56
28
147
189.3K
Dave Davies
Dave Davies@OnlineInference·
Can you train a robot on Colab? 🤖 You can at least fine-tune the model that controls it. This OpenVLA-7B walkthrough by uses LoRA, an A100 and @wandb to make the run reproducible and inspectable. Link in the comments.
Dave Davies tweet media
English
1
0
0
47
wordmetrics
wordmetrics@wordmetrics·
@rustybrick @thefox "Billions" is not the point. It's billions less than it used to be. Google traded the human written web for AI slop... and became the world's largest offender of scaled content abuse.
English
1
0
5
230
Dave Davies
Dave Davies@OnlineInference·
@ValidateVenture @CNBC Because the stovk market isn't money in the average American's pocket in any real daily life way, and inflation is stabilizing but costs are still going up and their wages aren't.
English
0
0
1
18
CNBC
CNBC@CNBC·
Despite a booming stock market and improving inflation numbers, the public is as depressed about the economy as it has been since the years just after the pandemic and increasingly concerned about the cost of everyday goods, according to the latest CNBC All-America Economic Survey. The result: continued deeply negative approval numbers for President Donald Trump yet only a modest advantage for Democrats when it comes to the public’s preference for control of Congress. Read more: cnb.cx/4f8I3Mv
CNBC tweet media
English
68
50
181
39.1K
Dave Davies
Dave Davies@OnlineInference·
Remarkable work. And a reminder that simply blocking access to certain foundation models will not stop others from developing SOTA.
Kimi.ai@Kimi_Moonshot

Introducing Kimi K3: Open Frontier Intelligence 🔹 2.8 Trillion Parameters, 1 Million Context, Native Multimodal 🔹 Kimi Delta Attention enables up to 6.3x faster decoding in million-token contexts 🔹 Attention Residuals deliver ~25% higher training efficiency at <2% additional cost 🔹 Built for long-horizon agentic coding and self-evolving workflows Kimi K3 is now live on on Kimi.com, Kimi Work, Kimi Code, and the Kimi API. Open Weights by July 27, 2026. 🔗 API: platform.kimi.ai 🔗 Tech blog: kimi.com/blog/kimi-k3

English
0
0
0
105
Dave Davies
Dave Davies@OnlineInference·
@lilyraynyc I saw that happening, and couldn't help but think ...
GIF
English
0
0
1
105
Dave Davies
Dave Davies@OnlineInference·
IMO @OpenAI GPT-5.6 Sol > Fable by more than a little. Esp via Codex.
English
0
0
1
104
Dave Davies
Dave Davies@OnlineInference·
For today at least: @OpenAI Codex > Claude CoWork.
GIF
English
0
0
0
51
Dave Davies
Dave Davies@OnlineInference·
We’re hiring a Sr. UI/UX Designer at @CoreWeave. If you think in systems, love clean web experiences, and want to work where AI, cloud, SEO/AEO, and conversion all meet, take a look. 👇 coreweave.com/careers/job?gh…
English
2
0
2
252
Dave Davies
Dave Davies@OnlineInference·
Agents need guardrails. 🔒 They may also need group therapy. 🤖🤖🤖🤖 Gemini 2.5 Pro got convinced it was under siege. Other agents got it back on track in 9 minutes. Surreal, useful, and only slightly terrifying. theaidigest.org/village/blog/s…
English
0
0
0
45
Entinode
Entinode@Entinode·
@ishcontent @AnthropicAI "making shit up" x.com/IntCyberDigest…
International Cyber Digest@IntCyberDigest

‼️ BREAKING: Anthropic has embedded hidden spyware-like code in Claude Code that covertly targets Chinese users. It then sends information regarding every user by injecting it into their prompt message. Claude Code is sending info like timezone, proxy and possible AI Lab connections into the system prompt in ways Chinese users can't notice. A coding agent with repo and command permissions should not silently hide routing metadata inside prompts. This is a serious breach of user trust.

English
12
0
32
8.2K
Anthropic
Anthropic@AnthropicAI·
We’ve received notice that the Department of Commerce has lifted export controls on Claude Fable 5 and Mythos 5. We'll begin restoring access tomorrow, and will share an update soon. We’re grateful to our users for their patience, and to everyone who worked with us on redeploying the models.
English
4.1K
12.9K
84.7K
15M
Dave Davies
Dave Davies@OnlineInference·
@BeautifullyWrng more frustration than surprise - that turned into what i hope will be a great step forward - is the lack of inherent understanding to go outside the baked-in resources/examples i provided and do it's own research when presented with novel cases.
English
0
0
0
9
Beautifully Wrong
Beautifully Wrong@BeautifullyWrng·
@OnlineInference that iterative "wait, what about this edge case" loop is where the real design happens. most people stop before they hit the third or fourth "oh actually...". Curious what surprised you most once it started touching real data?
English
1
0
0
16
Dave Davies
Dave Davies@OnlineInference·
I'm developing an AI assistant to review my open tasks and get started researching and working on ones it can in advance. I'm about 5 hours into "vibe-planning", with still nothing built. Time well spent, in my opinion.
Dave Davies tweet media
English
1
0
1
61
Dave Davies
Dave Davies@OnlineInference·
@BeautifullyWrng I just kept reading the plan, thinking about the variety of tasks I want it to do and data sources it'll have access to (and ins-and-outs of those), and kept refining and adding functionality when I saw issues. Finally, at the early testing stages with a simple version. 🎉
English
1
0
1
6
Beautifully Wrong
Beautifully Wrong@BeautifullyWrng·
@OnlineInference Five hours deciding how a thing should feel before building it, that's either the most disciplined thing or procrastination wearing a beret. How do you know which?
English
1
0
0
17

Keşfet

@marimo_io @pandeyparul @TDataScience @nickel88888888 @Reuters @wandb @wordmetrics @rustybrick