AI: A Bubble Waiting to Burst? Think Again!

Introduction

This post provides an overview of the ongoing debate regarding the notion that AI is a bubble on the verge of bursting. Various analyses and articles have compared the current AI landscape to the DotCom bubble. Notably, Allison Nathan, Jim Covello, and Daron Acemoglu of Goldman Sachs have expressed skepticism about AI's return on investment, detailed in their report, "A Skeptical Look at AI Investment." In contrast, David Solomon of Goldman Sachs does not share this view and recently hosted an interview with Eric Schmidt on AI's potential, highlighting several promising use cases. The full interview can be watched on YouTube.

This divergence of opinions within the same company is fascinating: Is AI a bubble, or is it truly revolutionary? With these mixed messages, I would like to offer my perspective. However, before doing so, I want to highlight some key points made by Eric Schmidt. In the YouTube video, Schmidt discusses AI's application in the financial sector, detailing various use cases. He emphasizes that financial institutions are primarily focused on increasing revenue. Schmidt explains how AI can drive revenue growth through enhanced product offerings, targeted messaging, and demand creation. He uses the example of AI optimizing advertisements to ensure they perfectly target the intended audience, stating, "The idea is to have AI and humans work together to drive revenue."

Schmidt envisions a future where AI is crucial for driving growth and revenue. He points out that developers will continue to write software, but AI models excel at writing code, underscored by Microsoft's substantial $13 billion investment in AI to assist developers. This investment aims to enhance developer productivity, generate more products, and ultimately increase profitability.

Diverging Perspectives

I hold a different view from the one presented in "A Skeptical Look at AI Investment." This post will outline why I believe the Goldman Sachs team is mistaken. We are at an inflection point where AI and machine learning (ML) represent an evolution in software development, capability, resulting use cases, and usage. Over time, AI and ML will integrate deeply into software for delivering real-world applications. It will become indistinguishable if the software users use their AI and machine learning capabilities. And this is the key today: if you perform a survey on people using ChatGPOT daily for work, it may not be good, but this is never how we are to use AI; this is just how it was provided on days one, but is not how we will be using AI and machine learning five years from now. We will see capabilities that would not have been possible just by using CPUs on their own. One key point is that achieving possible functional or machine learning results is only possible within the boundaries and architecture of existing CPU architectures.

Software Evolution

The software landscape is changing. Applications are no longer confined to traditional CPU architectures like x86 and ARM. Our applications are evolving to target new types of use cases. AI and ML, often considered separate from traditional software, are simply part of this software evolution, driven by the need to solve real-world problems. The potential applications are vast, and I have included a list of some use cases at the end of this post.

Overcoming Traditional Constraints

Historically, software applications were designed to run exclusively on CPUs, which imposed certain limitations. Despite advances in CPU speed and memory, these constraints shaped our capabilities. Enhancing CPUs with new capabilities to perform AI and ML functions opens up possibilities for solving everyday problems on a global scale. This is happening now; we are augmenting CPUs with functions for AI and ML, creating software that can address issues previously unsolvable with traditional CPUs.

Integrating AI with CPUs

Integrating AI-related functions with CPUs is a significant leap forward. It expands software capabilities to tackle new and uncharted use cases. The software industry is exploring these capabilities to find innovative ways to use them profitably.

Emerging Capabilities

As these enhanced capabilities emerge, we see the development of new types of applications. Just as a child learns to build complex structures with Lego, developers are mastering the creation of sophisticated applications using AI and ML functionalities.

Historical Context and Future Trends

The history of CPU advancements shows that expanding capabilities consistently leads to new possibilities and solutions. This trend continues with AI and ML integration, promising groundbreaking applications and solutions in the future.

Infrastructure Investments

The infrastructure to run AI and ML at scale must be improved globally. Major cloud providers like Microsoft have invested billions in building the necessary infrastructure, such as supercomputers for AI training. This investment is essential for delivering new applications with AI and ML capabilities.

Critique of the Goldman Sachs Report

The Goldman Sachs report argues that there are no use cases for AI and ML that will return the investment. However, cloud providers have a track record of successful infrastructure investments. Capacity management is critical for them, ensuring that assets are never idle. Given their expertise, I have confidence in cloud providers, and they are betting on AI and machine learning applications for the future. And it's hard to see why not when we start to look at future software, adding new AI and machine learning capabilities to solve problems. It's like saying we stop investing in software full stop.

The proof is in history; today, cloud providers have built out cloud services globally for delivering software and data, and AI and machine learning are extensions of this delivery. Just look at the sea of software and data services available from cloud providers to enable software and data delivery. Why would one think that if you can extend only a percentage of these with AI, how can you add value? Each software application running in the public cloud and on-prem can have value added to it with AI and Machine learning; it is like a gold mine, and raising the value of the application is the journey.

Use Cases for AI and Machine Learning

Fraud Detection

AI can analyze vast amounts of transaction data in real time to identify unusual patterns indicative of fraud. Machine learning algorithms continuously improve to counter new fraud tactics, enhancing protection and customer trust.

Cybersecurity Defense

AI detects and responds to cyber threats in real time, identifying patterns in network traffic and predicting future attacks based on past data. This proactive approach helps protect sensitive data and maintain system integrity.

DNA Therapies

AI accelerates the development of personalized DNA therapies by analyzing genetic data, predicting treatment responses, and optimizing gene editing techniques.

Drug Discovery

AI revolutionizes drug discovery by screening chemical compounds, predicting biological interactions, and optimizing drug formulations, leading to more effective medications.

Healthcare Delivery

AI improves diagnostic accuracy through imaging analysis and predictive analytics, optimizes treatment plans, and provides virtual health assistance, reducing the burden on healthcare professionals.

Service Industry

AI automates routine tasks, improves efficiency, and reduces costs. In customer service, AI chatbots handle inquiries 24/7, while in logistics, AI optimizes routes and inventory management.

Driverless Cars

AI enables autonomous vehicles to navigate complex environments, detect obstacles, predict traffic patterns, and ensure safety, promising to reduce accidents and transform transportation.

Personalized Marketing

AI analyzes consumer behavior to deliver targeted marketing campaigns, increasing engagement and conversion rates.

Film and Media Creation

AI assists in scriptwriting, storyboarding, special effects, and post-production, analyzing audience preferences to tailor content and streamline editing processes.

Music Creation

AI composes music tailored to individual tastes, analyzes musical trends, and generates compositions in various styles, aiding musicians and producers.

Video Editing

AI automates cutting, trimming, and color correction, enhancing footage and streamlining editing.

Smart Homes

AI automates lighting, heating, cooling, and security systems to optimize energy usage and monitor for intruders in real time.

Education and Learning

AI-powered educational platforms provide personalized lessons and feedback, ensuring students progress at their own pace.

Agriculture Optimization

AI monitors soil conditions, weather patterns, and crop health, enabling data-driven decisions to enhance productivity and sustainability.

Conclusion

AI and ML are integral to the future of software development, offering solutions across various industries. The investment in AI infrastructure is justified by the potential for innovative applications and the transformation of existing processes. Like all technologies, we are going to see some companies invest efforts in the wrong areas, and AI is no different; in the scramble, it may create businesses that fail to have real-world use, and we will also see organizations that do not fully understand where to focus efforts in AI to be successful, and these will also fail. This is not due to AI. Over the coming years, we will see AI and machine learning opportunities blossom; we have to take it steady as we go and ensure that we do not get pulled into all the hype. Creating a chance to generate any business needs a proper foundation, AI or not, and this does not stop because of AI.