GEN:48 submissions are now closed. All finalists will be revealed on 10/27 at 9 am EST for public voting. If you haven’t already, time to start working on submissions for AIFF 2024.
GENERATIVE AI
-
Auto-Regressive LLMs Limited Reasoning Cannot Be Fixed by Scaling
By
–
Anyone who thinks Auto-Regressive LLMs are getting close to human-level AI, or merely need to be scaled up to get there, *must* read this. AR-LLMs have very limited reasoning and planning abilities.
This will not be fixed by making them bigger and training them on more data. -
EY.ai Workforce Platform Streamlines HR Tasks with IBM Watson
By
–
.
@HRExecMag examines "
@EY
.ai Workforce", a new platfrom powered by @IBM
’s watsonx Orchestrate designed to help #HR employees streamline their work, leveraging #AI, #NLP and #automation to create job descriptions and payroll reports: https://
ibm.co/3Q49gCp -
Replicate Remove-BG Tool Alternative Recommendation
By
–
For folks seeing this issue, I recommend switching to https://
replicate.com/lucataco/remov
e-bg
… for now. -
Uncensored AI, essential fight of the next 10 years
By
–
In 10 years, we will look back and realize we should have been fighting for uncensored AI this whole time.
-
DALLE3 Creates Futuristic Green-Tech Vision Barcelona
By
–
A futuristic green-tech vision of Barcelona by DALLE3
-
GPT-4 DALLE3 prompt interpretation differences explained
By
–
And as pointed out, if you are on the web-version of GPT4/DALLE3 you can indeed click on the images and get your full prompts, but these are still not the ones you wrote in the first place. But the GPT4 interpretation based on your initial ask.
-
Extracting GPT-4 System Instructions: Challenges and Insights
By
–
Here is how the pre-cursor instructions are, I've yet been able to get GPT4 to exactly spill the beans on the system instructions.
-
Closed Frontier Labs Cannot Maintain Competitive Advantage Long Term
By
–
Major fallacy here: "closed frontier labs" can't possibly stay simultaneously "closed" and "frontier" for very long.
They must either open up or fall behind.
In any case, they will lose whatever advance they think they have.
Why? Because people move around and ideas get around. -
Emergent Abilities in AI: Training’s Hidden Role Revealed
By
–
I’ve often thought this. The “emergent” abilities seem less emergent when you know, for example, that skilled programmers have spent thousands of hours training the model.