🧠 Is super-intelligence legit?

If you’re reading this, there is a chance you think lots of businesses today use AI. I thought so too…

The fact is that while AI hype is soaring, implementation is lagging. 

Despite half of the S&P 500 companies discussing AI as frequently as “interest rates” during earnings calls, a recent NBC News analysis revealed that only 4.4% of businesses nationwide actively use AI for goods or services. 

Even with the surge in AI interest courtesy of OpenAI's ChatGPT, the talk-to-action ratio remains skewed.

Why the divide? One of the biggest reasons is that the tech is still too nascent, contributing to the cautious approach.

Additionally, a study by Kristina McElheran at the University of Toronto highlights a resource gap, with expensive tools, limited expertise, and insufficient education all hindering widespread AI adoption.

Overall: AI is an important and influential technology, but businesses are still reluctant to use it in their work.

As businesses navigate this landscape, they are likely to adopt user-friendly applications built atop AI for smoother and more efficient operations instead of directly engaging with AI.

Read more: NBC News

In a recent interview, Meta’s AI Chief, Yann LeCun, had some interesting thoughts on the development of AI.

On Super-Intelligence:

Are we on the brink of achieving AI super-intelligence?

Contrary to Nvidia's CEO, who anticipates AI reaching human competitiveness within five years, LeCun believes we are still decades away from attaining AI with human-like common sense.

LeCun dismisses the notion of imminent super-intelligence, suggesting that the initial milestones will likely manifest as "cat-level" or "dog-level" AI.

On AI Models

When talking about the evolution of AI systems, he advocates for a shift beyond text-centric models, stressing the importance of embracing multimodal AI that seamlessly integrates audio, image, and video data.

It makes sense, given Meta’s all-in mentality on combining AI with AR and VR. 

On AI Chips:

LeCun's vision extends to advancements in AI chips, transcending their traditional graphical nature. He envisions a new generation of neural, deep-learning accelerators that optimize matrix and vector operations - fundamental components of neural network computations.

In simpler terms, these accelerators will be designed to tackle complex mathematical operations more efficiently, surpassing the capabilities of general-purpose processors.

Exciting times ahead for the development of AI. 

Read more: CNBC

The US military is looking to ramp up its involvement with AI.

Air Force Secretary Frank Kendall recently spoke about the urgent need for the U.S. military to swiftly adopt AI for enhanced target tracking, data analysis, and mission management. 

When questioned about AI’s unintended consequences, he stated that AI's role is in augmenting human intelligence rather than fueling an arms race. He also assured compliance with international law, emphasizing a human-in-the-loop approach for all lethal decisions. 

Overall: The military envisions AI as integral to battle management in space, leveraging Earth observation data for strategic insights. 

They urge substantial investment in AI resources and real products to harness its potential for automation and efficiency, but the real concerns will always be around their ethical and legal standards.

What do you think?

Read more: Space News

What else is going on?

This look into Figure’s AI Robotics lab is quite fascinating 🦾