Guido Appenzeller said Anthropic defines an agent as an LLM running in a loop with tool use.
“a another definition we heard from Anthropic recent recently was this idea that an agent is an LLM running in a loop with tool use, right?”
Person
Guido Appenzeller
1 episodes · 7 quoted claims
Guido Appenzeller described a startup pitch pricing an agent at about $30,000 a year against a human worker earning about $50,000.
“The human worker makes, I don't know, 50,000 a year, and therefore this agent, you can get for only 30,000 a year.”
Guido Appenzeller argued that product prices tend to converge toward the marginal cost of production over time.
“we all know that the cost of a product over time converges towards the marginal cost of production, right?”
Guido Appenzeller said LLMs need large GPU clusters and are typically run on separate specialized infrastructure from the agent's own logic.
“you need these vast GPU farms, uh, can't easily run today's LLMs on a single GPU, so that's that's a very specialized infrastructure that's externally.”
Guido Appenzeller cited iPhone photos as data not accessible through any API because of a walled garden.
“My iPhone, the photos are not accessible via any API because it's a walled garden,”
Guido Appenzeller said a deep research tool from a major LLM provider reasoned about getting around a captcha and succeeded.
“one of the steps, if you look through it, all the steps it went through was like, you know, trying to see how I can get get around a capture mechanism for a site.”
Guido Appenzeller said security, authentication and access control for agents acting on a user's behalf remain unsolved.
“We have not figured out security, authentication, access control for agents working on my behalf yet, right?”