you.com is a productivity engine. Millions of people rely on our AI Assistants to conduct detailed research and analysis, solve complex problems, and create personalized content.
Businesses trust our platform to maximize the full potential of their employees and products.
Focus is crucial for highly impactful work, but focus is often compared to putting blinders on or otherwise constraining and narrowing vision.
This is not the right metaphor for focus.
True focus allows you to see more not less. It brings the right questions, considerations, and solutions into the foreground.
Only by studying philosophy did I figure out what I wanted to do with computer science. Only by studying computer science did I find the methods for approaching the philosophical questions that concerned me.
It was a very particular kind of focus that drove me to cross and connect fields to find people like @RichardSocher @youdotcom working on the right questions in the right ways at the right time. #AI
As LLMs support more languages, especially under-resourced ones, startups like you.com will benefit from a global user base.
At you.com, we collaborate with companies like Upstage to build amazing experiences for the Korean and Japanese markets.
We want everyone to experience our productivity engine.
The next level in AI will come through simulations.
Anything you can simulate, you can eventually solve with AI since you can create unlimited training data. This is already true for games like Go, where the AI thought of new moves that humans didn't come up with.
It will start with simple physical simulations, programming/coding environments, and, eventually, more complex biological systems like a cell.
It will be able to solve so many fundamental problems and turn scientific fields into engineering disciplines.
The future of engineering in the age of agents and autopilots is a leading indicator for all knowledge work, and begs the question...
Who are we really building AI agents for?
Earlier this week, I discussed what might happen to entry-level engineers and suggested engineering education might evolve like a medical residency. But the implications go well beyond engineers.
Will tech giants dominate further, or will AI agents enable more disruptive startups?
What role can open source play in determining the answers to these questions?
If you'd like to explore these questions together at #SXSW, vote for our session on exactly these topics!
You can vote here: https://lnkd.in/gdU2zByW
The next level in AI will come through simulations.
Anything you can simulate, you can eventually solve with AI since you can create unlimited training data. This is already true for games like Go, where the AI thought of new moves that humans didn't come up with.
It will start with simple physical simulations, programming/coding environments, and, eventually, more complex biological systems like a cell.
It will be able to solve so many fundamental problems and turn scientific fields into engineering disciplines.
Knowledge workers are quickly adopting AI Assistants for everyday tasks that require accurate, up-to-date answers.
We're already building agent-like features for research, analysis, problem-solving, and content creation. The better LLMs become, the better these abilities become.
We've also layered on an index of the web purpose-built for LLMs with more relevant information per page that can be incorporated with Retrieval-augmented Generation (RAG) over company data.
The pendulum swings back and forth between large and small AI models.
One of the most fascinating parts of watching AI go mainstream has been watching an ever-increasing number of people see the pendulum swing for the first time.
We should all be rightly excited as small models improve, but it is just a matter of time before an even bigger, more unified model sets a new record for blowing our minds.
Then we'll spend some time as a community distilling (broadly speaking, not only in the precise technical way that word is used) all the learnings that went into reaching that new high point into smaller models again.
Bigger, larger models are better for searching for new, mind-blowing capabilities.
All those parameters help us search for new solutions.
But once we've found those new solutions, we can package them up into smaller models that only need to execute on that solution or serve as a base for a comparatively small optimization for a particular application.
In reality, the pendulum swinging is an illusion -- it reflects the community's attention on either small or large models as the immediate next step.
But AI advances in bipedal strides -- left foot small models, right foot big models.
For those always searching for the new solutions there is a constant stream of effort being poured into the main line of research, driving that right foot forward while the left is planted.
For those always searching to apply the latest and greatest to their applications and products, there is an equally constant stream of effort, driving the left forward once the right is planted yet again.
And so we advance step by step...
#AI
Knowledge workers are quickly adopting AI Assistants for everyday tasks that require accurate, up-to-date answers.
We're already building agent-like features for research, analysis, problem-solving, and content creation. The better LLMs become, the better these abilities become.
We've also layered on an index of the web purpose-built for LLMs with more relevant information per page that can be incorporated with Retrieval-augmented Generation (RAG) over company data.
If you'd like to learn how to make LLMs more useful and trustworthy, vote for my #SXSW 2025 session proposal —"Beyond Proof-of-Concept: Making AI Useful For Business"
Voting ends Aug. 18 🤙
https://lnkd.in/gZ-9PM9x
As engineering copilots become autopilots, all of the entry-level work software engineers are doing today will be automated with AI.
Consequently, all engineers will be AI engineers in the future, in a much more literal sense than it is often meant today.
Future entry-level engineers will need to learn something quite different, focusing on engineering at a higher level of abstraction.
Something akin to medical residencies may become the norm for AI and engineering.
Engineers will no longer ever write code for a task directly.
They will instead solely engineer the AI that writes the code that engineers do today.
All engineering will have become AI engineering.
For those that want to truly specialize in an area that is 99.99% automated and optimized by AI, there may be a long period of time and training followed by a quick jump to extremely value -- at the point when the specialist can make the AI better, a kind of AI neurosurgeon.
Similar lines of reasoning apply to all jobs in knowledge work.
#AI