What news from AWS re:Invent last week will have the most impact on you?
Amazon Q, an AI chatbot for explaining how AWS works.
Super-fast S3 Express storage.
New Graviton 4 processor instances.
Emily Freeman leaving AWS.
I don't use AWS, so none of this will affect me.
AI / Data

The AI Evolution Will Happen Faster Than Computing Evolution

We now face our responsibility with AI. Will we act responsibly or let it become something that we profoundly regret for how it affects our humanity?
Oct 20th, 2023 12:58pm by
Featued image for: The AI Evolution Will Happen Faster Than Computing Evolution
Feature image by Tumisu from Pixabay.

The AI evolution will happen faster than the computing evolution. That’s how Raj Verma, CEO of SingleStore painted the current state of matters at the SingleStore Now conference in San Francisco this week.

A host of factors play into this scenario but with it all comes the need for responsibility and what might be evoked when we view the context of a work by Michelangelo.

Verma used the “Creation of Adam” painted by Michelangelo in the 16th century to show the symbolic representation of man and AI. The human’s ability to attain intelligence and wisdom correlates to the difference in how we consider computing power and AI systems’ role. The brain is our infrastructure. That’s not what Verma stated directly but it holds up when thinking through how humanity now depends on software and increasingly AI to augment our own intelligence. Our brains have evolved incrementally just as compute and infrastructure have over the past several decades. We’ve built mainframes, personal computers, and devices, most of all the iPhone, which Verma views as the most significant development in leading us to where we sit today.

Compute will still evolve, but how fast is the question? The Internet led to the massively distributed data center approach, which we know as the cloud, a terrible term, but I digress. But today the power of computing can only increase so much. Moore’s Law looks increasingly impossible to keep pace with as we develop transistors the size of an atom.  Infrastructure limitations are causing all sorts of headaches for software vendors who now face a litany of options for maximizing AI systems to be more efficient with precious compute resources.

And that brings us back to Verma, his imagery of Michelangelo and why it’s data that we must consider as the true catalyst in understanding why AI will evolve faster than compute.

With AI systems, each advancement becomes the foundation for unlocking new capabilities. There is also the possibility of self-reinforcing capabilities that, in turn, will speed progress. AI progress is exponential. Breakthroughs mean new advancements.

But what do we value most? The intelligence that comes when we use pre-trained models; the groundbreaking advancements in AI when we use foundation models. It’s the data that matters. And if does not come fast then forget real-time apps. They just won’t work. And who the hell wants a slow experience? It’s just not realistic to expect any advancements without fast data.

The SingleStore Angle

SingleStore, formerly MemSQL, started as a database company in 2011. The company’s history tells a story of economics and opportunity that turned the company’s focus from in-memory SQL. They made a bet in those early days: the price of memory would decrease enough for performance to reside entirely in memory.

The prices did not fall, but they had the right idea. Data volumes continued to increase in size. So the team decided to add disk to their in-memory capabilities, resulting in a hybrid architecture that set them up to make fast data accessible in-memory and the cold data on disk. Users could then get real-time transactional data and real-time analytics on the full data set. In 2020, the company adopted the SingleStore brand name.

How does this relate to AI’s evolution? It’s all about the data and its compounding growth. Having transition data ready and analytic, too, with speed and efficiency, makes for the ability to scale AI systems. As we’ve seen, AI systems must be fast and SingleStore markets that capability with its in-memory and disk capabilities. There’s also the flexibility that customers demand — a hybrid approach that cuts across cloud services and on-premises.

With SinglStore vector indexing and JSON handling, the capabilities opened further. The ease of ingesting data, normalization, and low latency are just some of the upsides of JSON in data store environments and AI systems. JSON  fits well with SQL and vectoring.

AI systems can only evolve with data. It’s very much akin to the evolution of humanity, which has evolved by absorbing data almost constantly. The pace of AI’s evolution will only get faster with more data.

Silicon Valley Meets Michelangelo

I can attest to sitting through many keynotes that come with self-serving declarations. But rare is the keynote that looks to Michelangelo for inspiration.

The concept of humanity in tech conversations seems as accelerated as the advancement of AI. But I like how Verma approached the matter in his address.

He reflected on Michelangelo’s Creation of Adam in the Sistine Chapel. It shows God imparting life to Adam, symbolizing the dawn of intelligence and wisdom.

In the past 50 years, art history scholars have viewed the Michelangelo painting as showing God in the context of the imagery of the brain, Verma said. The painting also shows God with his arm around Sophia, the ancient Greek goddess of wisdom. And there we see it. The brain as our core architecture for absorbing data, thinking through the context and developing intelligence.

What do we see with AI? We are trying to mimic this ability we see in the painting by giving intellect and wisdom to machines by delivering data fast, acting on it, and using the data store for analytics. The speed of in-memory and disk becomes a way for developers to work with services that offer developers a platform for building real-time applications.

For Verma, the painting tells another story — our journey to create machines that possess their own intelligence. It’s a view that speaks to the driver in recreating aspects of cognition through AI.

San Francisco is a tech-positive city. Rarely do we hear a CEO speak to the negative implications of technology and how we use it.

Yes, the brain is evident as God touches Adam in the painting. We know well the negative implications that come with humanity that we must consider when viewing Michelangelo’s creation. Like the power of our intellect, AI has potential negative impacts if used irresponsibly — case in point — social media.

Verma said that social media connects us, but the downsides are sadly apparent in the impact on teenage girls.

“I can text my loved ones and know where they are and their locations and all the rest of it and helps me sometimes sleep better at night,” Verma said.

But at the same time, there have been disturbing downsides. For example, Verma cited increased teenage suicides, especially in girls. Contributing to the increase — the awful practice of cyberbullying and the unhealthy focus on body image that is so damaging to girls.

The giant social networks? How fast have they reacted to the issue? Facebook considers itself a printing press and has “hidden behind that chestnut,” Verma said.

We now face our responsibility with AI. Will we act responsibly or let it become something that we profoundly regret for how it affects our humanity? How fast is too fast?

Group Created with Sketch.
TNS owner Insight Partners is an investor in: Pragma, SingleStore.
THE NEW STACK UPDATE A newsletter digest of the week’s most important stories & analyses.