Very large context windows may extend the capabilities of LLMs because you can give them hundreds of examples on how to solve a problem (many shot learning). This paper from Google finds significant performance boosts from many shot, even when the AI generates its own examples.
@emollick
-
Employees as R&D: Frontier AI Access and Experimentation Strategy
By
–
Your R&D team for using AI is your employees, they can better experiment and judge the results of AI co-intelligence than anyone in a central research organizations. Paying for frontier AI access for employees, and giving people time and incentives to experiment, is an R&D cost.
-
Controlled AI Experiments Show Real-World Abilities Across Fields
By
–
So much debate over the utility of AI is based on vibes alone, most doesn't engage with the increasing number of controlled experiments that show strong real world abilities of AI across a variety of fields. Implementation takes time, of course, but this isn't just speculation.
-
GPT-4 adoption in business shows no skepticism from users
By
–
Its all anecdotal, of course, but not seeing any skepticism that this is a big deal from organizations actually using GPT-4 in businesses.
-
MIT CEOs Report Significant AI Performance Impact
By
–
At a conference at MIT, talking with a range of CEOs, technical leads in firms, etc. These are mostly people ahead of the curve in experimenting with AI in their organization. Every one of the ones applying it was saying it had a big impact on performance in their organization.
-
Software Systems Design: Beyond Explicit Function Engineering
By
–
I think many people are used to thinking of software systems as something where all the functions are explicitly designed.
-
Apple’s Spatial Computing Lacks Imagination Despite 3D Portal Potential
By
–
The most disappointing thing is ultimately that Apple made every single experience so boring. You can create 3D portals to anywhere and they decided the only thing people want to do is to create multiple virtual 2D computer monitors. Amazing possibilities, but no imagination.
-
LLMs Develop Unexpected Retrieval Heads for Context Windows
By
–
Fascinating thread in two ways. The first is the paper and findings itself, explaining how LLMs actually deal with context windows. The second is the fact that this is a “discovery:” LLMs have unexpectedly developed retrieval heads, they were not explicitly coded for by creators
-
AI Won’t Take Jobs, It Will Transform Them
By
–
I see “AI won’t take your job, someone using AI will take your job.“ don’t like this frame because:
1) “Your job”’ is more likely to transform over time than be “taken”
2) Some jobs will really disappear with AI
3) Using AI is going to get easier, it is isn’t a secret priesthood -
Copilot’s Impact on Coding Productivity Measured
By
–
Field experiments are great but they can go wrong in so many ways because the real world is messy I appreciate this honest write-up of the difficulties of measuring the impact of Copilot on coding performance. It looks like it lifted productivity, though. https://
mit-genai.pubpub.org/pub/v5iixksv/r
elease/2?readingCollection=7e70f4fe
…