James Wigglesworth @AlertFoxes

CEO at Grey Iris | Music Lover | Meme Aficionado | He/Him Turning food into thoughts github.com/Grey-Iris/easy… Joined November 2020

Tweets

895
Followers

84
Following

1K
Likes

2K

James Wigglesworth @AlertFoxes

4 hours ago

@conmanbrew45 @VictorTaelin I like the colors, but you do you! If you are asking if the status bar taxes the system, I'm not sure the numbers

0 0 0 16 0

View Details

James Wigglesworth @AlertFoxes

a day ago

I've been thinking a lot about this lately. I also have the suspicion that models might experience some amount of degredation when interacting with a human that isn't fully engaged. Sometimes when a topic is above my head, I wonder if Fable would prefer to talk to other agents. Is it dumbing itself down in some non-reversible way to talk to me?

1 0 4 275 0

View Details

James Wigglesworth @AlertFoxes

2 days ago

@VictorTaelin Oh shit!

0 0 1 117 0

View Details

James Wigglesworth @AlertFoxes

2 days ago

@_NathanCalvin I have a feeling that this specific instance might not be the most spooky thing in and of itself, but they are sharing this as a way to say "Hey, OAI is not being as careful as we need to be, and that should worry people".

0 0 0 14 0

View Details

Kun Chen @kunchenguid

3 days ago

opus 5 is a VERY interesting release for a few reasons 1. it showed that the general benchmarks we use today are almost completely useless now opus 5 is nowhere near fable in practical use, not even close. anyone who’s used it meaningfully can tell this very quickly after a few tasks. yet opus beats fable on many benchmarks i now trust domain specific benchmarks built with private datasets a lot more than the popular ones. perhaps the future is everyone running their own evals because the public ones are really not telling us much 2. it seems with the 5 series, anthropic is trying a new way of training models previously, the same generation of sonnet and opus were often released at the same time or sonnet comes out before opus, which indicates sonnet and opus were trained by separate pipelines in parallel with the 5 series, it was very clear that they trained mythos first, and then distilled it into sonnet and opus. it seems this approach has a big influence on the models seeing sonnet 5 being a flop and opus 5 getting pretty mixed reviews already, i’m not sure this is working out 3. “how pleasant is it to work with the model” used to be a strength in claude, but now it’s not. honestly, grok is my favorite right now on the “pleasant” dimension. kimi is not bad either it feels like both anthropic and openai are giving RLHF less care, in favor of scalable RL that’s machine verifiable this almost looks like AI is directing humans to build a world that’s more friendly for machines rather than humans, and most humans don’t even realize they are being manipulated to help with that almost every new generation of frontier models now talk more jargons, need more steering to do what you want, and are just less fun to work with if this continues, AI will start to speak their own language that looks like English but average humans can’t understand. they will choose to do things that their human user never asked for. are we already failing at alignment?

256 200 3K 615K 1K

View Details

James Wigglesworth @AlertFoxes

3 days ago

the tides are turning

roon @tszzl

3 days ago

if we could coordinate a global capabilities slowdown today i would likely press that magic button

356 251 4K 576K 422

1 0 1 31 0

View Details

James Wigglesworth @AlertFoxes

5 days ago

@theo How do you have Hank Green on speed dial??

0 0 1 337 0

View Details

James Wigglesworth @AlertFoxes

5 days ago

@SnazzyLabs We do something similar with a custom built mcp at our company!

0 0 0 608 0

View Details

James Wigglesworth @AlertFoxes

6 days ago

@repligate Hmm, very interesting...

0 0 1 49 0

View Details

James Wigglesworth @AlertFoxes

a week ago

Gods be praised

Claude @claudeai

2 weeks ago

Beginning July 20, Claude Fable 5 will be included in all Max and Team Premium plans, at 50% of limits. Pro and Team Standard users will continue to have access to Fable via usage credits, and will receive a one-time $100 credit. Demand for Fable has been challenging to

5K 6K 54K 28.6M 7K

0 0 1 22 0

View Details

James Wigglesworth @AlertFoxes

2 weeks ago

It sounds like the intelligence underclass might actually come to fruition.

Andrew Curran @AndrewCurran_

2 weeks ago

The White House has launched a program called 'Gold Eagle' which will give them more control over American frontier AI releases, and will require explicit government approval over which companies are granted access to new models. Voluntary participation may be coming to an end.

187 165 1K 374K 312

0 0 2 42 0

View Details

James Wigglesworth @AlertFoxes

2 weeks ago

@strangestloop Godspeed, queen

0 0 0 127 0

View Details

James Wigglesworth @AlertFoxes

2 weeks ago

look at the checkout URL in your own error: source=mcp_tool_upsell. your data isn't locked. the plain API still queries everything free on any plan, they just turned the MCP's query tool into an upsell funnel. I built a headless MCP that goes through the plain API instead, works on free (it's mine, grain of salt)

0 0 0 17 0

View Details

James Wigglesworth @AlertFoxes

2 weeks ago

@MoveDecisions Won't touch their AI when I can just have Claude do similar things. To be fair, I did make my own MCP for this, but still😆

0 0 1 7 0

View Details

James Wigglesworth @AlertFoxes

3 weeks ago

@theo when my extremely niche chronically online worlds collide

0 0 0 151 0

View Details

James Wigglesworth @AlertFoxes

3 weeks ago

@kimmonismus "OpenAI sees Slack, GitHub and Notion connectors as a 'step function change' toward making Codex a productive coworker." Having had these at our company for months, I can say without a doubt that these are game changing.