
I push the pretty pixels ๐๏ธโจ I lift heavy circles ๐๏ธโโ๏ธ I make games at @EmbarkStudios ๐ฉโ๐ป she/her
Stockholm, Sweden
Joined March 2017
- Tweets22.3K
- Following1.2K
- Followers2.7K
- Likes63.6K
Pinned Tweet
I posted these recently but it's time for a new pinned tweet, so...
Hi, I'm Gray! I'm an illustrator and technical artist that loves to draw badass women and use programming to create & aid my art.
๐ Email: [email protected]
๐๏ธ More: grayolson.artstation.com/
gray ๐ retweeted
This is my experience too. It's a huge, obvious effect. Now put this alongside the utterly confident claims that "AI writes better code than you ever did." xitter.cf/Sirupsen/status/211290โฆ
Replying to @CarlosPitteri @CDoombeard
This is way worse than harvesting. They could for example immitate youtube on your TV, steal the gmail password you log in with, and turn that into a password reset or verification vector for anything: a SIM connected to ur phone #, combine those for a 2FA direct bank login โ at Stockholm, Sweden
This has been an ambition since basically day 1, and I fondly remember watching many fun prototypes at our internal "show & tell"s, but taking this kind of feature from prototype to something that can actually ship is a monumental challenge. So cool it's finally here!
gray ๐ retweeted
I'm simply not going to have it output 100k loc
Code is literally useless. Code in fact is a liability. Having more code is actually worse overall. I am trying to solve problems with the least amount of code possible. This is very, very important
gray ๐ retweeted
"AI can replace artists" โperson whose walls are empty other than a superhero movie poster
"AI can replace musicians" โperson who never goes to live performances
"AI can replace authors" โperson who only reads Twitter
both cute and a wee bit terrifying
Anthropic's Frontier Red Team believes small AI models may now be months away from being capable of self replicating without access to a datacenter.
Quoting from the blog:
'The capabilities that make AI models generally powerful coding agents also generalize to cyber tasks, including autonomous replication. Like coding agents in 2024, weโre seeing signs of life for AI worms. We donโt yet have a full rogue replicator, but agents are becoming increasingly capable of completing the individual steps in the replication chain (and passing on their adaptations to future copies of themselves).
At the same time, increasing amounts of inference compute are now available both inside and outside of datacenters (see Appendix D). Itโs uncertain exactly what form autonomous AI replication will take. It could be a slow, distributed โdigital slime mold,โ living on relatively insecure compute. It could also be a powerful cyber model that breaches datacenters at a high pace, or a model that scams people for compute credits.
As a prediction that goes beyond our experiments, we expect a future awash with agents running on all types of hardware, including some that arenโt under human control. While itโs true that IoT devices have been vulnerable for a long time, lightweight agents may be much more willing than human attackers to put in the effort to compromise large fractions of the worldโs IoT compute. The rise of slow replicators would also be a bellwether for more aggressive, faster ones that can breach more hardened machines or be effective API key thieves. We should aim to harden our systems to make it difficult for such replicators to spread, but also prepare to deal with a world where they are here to stay.'
gray ๐ retweeted
It is true.
In particular, the x86_64 instruction set is a baroque fiction, itself implemented in microcode, WHICH CAN EVEN BE DYNAMICALLY UPDATED, on top of the actual processor cores that, trust me, you have not the first clue how to program.
























