weekend ai reads for 2024-02-23

Laika is an AI teenager raised solely on social media. Her personality, ideals, and opinions are 100 % shaped by the content and climate of popular digital platforms. Through her, we can observe and understand the potential risks of consuming too much social media, without exposing real kids. Laika is used for mental health research and education.

My Tester continues in an unnervingly human vein, promising me she will dress as Jet from Gladiators and serve me virtual pints during the Spurs game. These AI bots are already remarkably well engineered to mimic the thrills and jeopardy of human seduction.

"Love and sex with robots and holograms are an inevitable reality. They are great companions and capable of expressing empathy. Just as phones saved us from loneliness and filled the void in our lives, holograms as interactive presences in our homes can take it even further," she said.

It feels like in a lot of possible scenarios, nothing we could do to prepare the kids will particularly matter. Or what turns out to be helpful is so weird we can’t predict it well. So we’re just thinking about this for the possible futures where some skills matter, and we can predict them to some degree.

People are using AI tools to help with the mundane stuff. They may not shout it from the rooftops, they may feel uneasy about where this is heading, but if it’s shaving an hour off their working day, then what harm?

Welcome to the post-post-post-truth AI world. You know it’s not real. But you find yourself making out with bots and shoes. And there is more out there. ASMR is distracting you from a slightly higher priority: survive. The Future will be weird AF.

The LPU is designed to overcome the two LLM bottlenecks: compute density and memory bandwidth. An LPU has greater compute capacity than a GPU and CPU in regards to LLMs. This reduces the amount of time per word calculated, allowing sequences of text to be generated much faster. Additionally, eliminating external memory bottlenecks enables the LPU Inference Engine to deliver orders of magnitude better performance on LLMs compared to GPUs.

Our experiment suggests that existing evidence of decreasing differences based on the routine use of AI advice to improve writing or the gathering of information may not translate easily to sectors such as medicine or engineering where subjects have expertise and may want to trust it more at the moment of the decision than any advice they receive.

It seems like a long way to go from videos of cute paper craft turtles to The Matrix, but it’s worth taking the progress they’ve made here seriously. In generating video, Sora does a few things that are really difficult, and resemble human developmental milestones.

Instead, it will reshape the value and nature of human expertise. Defining terms, expertise refers to the knowledge or competency required to accomplish a particular task like taking vital signs, coding an app or catering a meal. Expertise commands a market premium if it is both necessary for accomplishing an objective and relatively scarce.

States could have much to gain from using generative AI tools in their operations. A low-risk strategy may be to test some applications while preparing a longer-term implementation plan.

It is unclear whether those included in the data consented to having their audio and texts used in such a manner. However, a sample dataset downloaded by The Daily Beast included a recording of a lecture from 1999 making it highly unlikely that they knew their data would be used to train future generative AI models.

In the desperate scramble to combat AI, there is a real danger of penalising students who have done nothing wrong

We can be critical of AI, but we can’t pretend it doesn’t exist if we want to prepare students for a world where humans will have to live and work alongside thinking machines.

Instant & personalized AI feedback for written assignments

This new generation also delivers a breakthrough in long-context understanding. We’ve been able to significantly increase the amount of information our models can process — running up to 1 million tokens consistently, achieving the longest context window of any large-scale foundation model yet.

As more developers begin to build using LLMs, however, we believe that this focus is rapidly changing: state-of-the-art AI results are increasingly obtained by compound systems with multiple components, not just monolithic models.

Meet Visual Electric—an image generator that's built for the creative process. Say goodbye to chat and say hello to the generative canvas.

If you have to choose between being nice and kind, the latter is a better option. The ultimate responsibility we all have is to be kind.