Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
2/ Agents will understand the context
“Claude needs to learn enough about your particular situation and the constraints you’re operating under to be useful. Things like what specific roles you have, what writing styles, or what needs you and your organization have.
ANTHROPICAL
“I think we’re going to see improvements there where Claude will be able to search things like your documents, your Slack, etc., and really learn what’s useful to you. This is a little understated by the agents. It is essential that the systems are not only useful but also secure, doing what you expect.
“Another thing is that a lot of the tasks won’t require much from Claude reasoning. You don’t have to sit and think for hours before opening Google Docs or something like that. And so I think a lot of what we’re going to see is not just more reasoning, but applying thinking when it’s really useful and important, but also not wasting time when it’s not necessary.”
3/ Agents will make coding assistants better
“We wanted to give developers an initial beta of using the computer to get feedback while the system was relatively primitive. But as these systems get better, they could be used more and really work with you on different activities.
“I think DoorDash, the browser company, and Canva are experimenting with different kinds of browser interactions and designing them with AI.
“I expect we’ll also see further improvements to the coding assistant. This is something that was very exciting for the developers. There’s just a ton of interest in using Claude 3.5 for coding, where it’s not just auto-completion like it was a few years ago. It’s really understanding what’s wrong with the code, debugging it—running the code, seeing what happens, and fixing it.”
4/ Agents will have to be safe
“We founded Anthropic because we expected artificial intelligence to advance very quickly and [thought] that security concerns will inevitably be relevant. And I think that will become more and more visceral this year, because I think these agents will become more and more integrated into the work that we do. We must be ready for challenges, such as a prompt injection.
[Prompt injection is an attack in which a malicious prompt is passed to a large language model in ways that its developers did not foresee or intend. One way to do this is to add the prompt to websites that models might visit.]
