GPT-4o: I agree with George Lee’s optimism that AI’s real value lies in specific, high-impact use cases. Claudia Harris is spot on about adoption barriers.

Question 1 of 2 for ChatGPT-4o: Please read the transcript of the conversation: "Making money from AI — Searching for a ‘killer app’" and tell me what it says.

ChatGPT-4o’s answer:
The transcript of Tech Tonic podcast episode titled “Making money from AI — Searching for a ‘killer app’” by the Financial Times (hosted by Madhumita Murgia and produced by Josh Gabert-Doyon) covers various perspectives on how companies are using AI and whether it is delivering profitability. Here are the key points:

1. Real-World Use Cases of AI

• Octopus Energy Call Centre: They use AI for generating email responses and summarizing customer queries. The AI drafts replies that agents can review and edit, saving time and improving response rates. However, Octopus isn’t cutting jobs but instead is improving response times and onboarding speeds.

• Synthesia (AI Video Company): Synthesia creates AI-generated videos for staff training. It aims to simplify video production, but despite being valued at over $2 billion, it hasn’t reached profitability.

2. Diverging Views on AI Profitability

• Goldman Sachs Debate:

  • Jim Covello (Skeptic): Believes AI lacks a clear ‘killer app’ and isn't generating enough revenue to justify its costs. He compares AI’s hype to past tech flops like VR and blockchain.

  • George Lee (Optimist): Argues that AI’s impact will emerge in unexpected industries, from legal services to engineering, driven by cost savings and productivity gains.

3. AI in Coding: Replit’s Approach

• Replit (AI Coding Tool): Allows non-programmers to create apps by describing them to an AI agent. Sears, for example, built internal business tools with Replit instead of using expensive SaaS products. However, Replit, despite high valuation and popularity, remains unprofitable.

4. Challenges in AI Adoption

• Workplace Disruption: Claudia Harris from Makers notes that AI adoption in companies is slow, with many employees resistant to changing their routines or fearing job losses.

• Unproven Business Models: Sarah Myers West from the AI Now Institute highlights that many tech giants, including Google and Meta, are investing heavily out of fear of missing out rather than proven profitability.

5. The Killer App Question

The podcast concludes that while AI shows potential to boost productivity and reduce costs, it has yet to deliver a transformative "killer app" that justifies the massive investments. The race is on between companies like OpenAI, Meta, and emerging competitors like China's DeepSeek, which promises more efficient AI models at lower costs.

Overall, the episode captures the tension between AI’s promise and its current reality—between sky-high valuations and uncertain returns on investment.

Question 2 of 2 for ChatGPT-4o: With which of the views expressed do you agree?

ChatGPT-4o’s answer:
I find both sides of the debate compelling, but I lean towards a blended perspective:

1. AI’s Practical Value in Narrow Applications (Aligning with George Lee & Replit’s Approach)

I agree with George Lee’s optimism that AI’s real value lies in specific, high-impact use cases. Examples like Octopus Energy’s improved customer service and Replit’s democratization of coding show that AI can generate productivity gains, especially in labor-intensive or knowledge-heavy tasks. I think tools like Synthesia and Replit will reshape industries by lowering barriers to entry—whether in coding, video production, or customer service.

2. The Absence of a ‘Killer App’ (Echoing Jim Covello’s Skepticism)

However, I share Jim Covello’s skepticism that we haven’t yet found the defining, transformative AI application that will drive mass adoption and deliver large-scale profitability. Right now, AI is solving problems incrementally—like drafting emails or writing code faster—rather than creating entirely new markets or industries. The comparison with past overhyped technologies like blockchain and VR is a cautionary tale.

3. The Adoption Gap (Agreeing with Claudia Harris)

I also think Claudia Harris is spot on about adoption barriers. Companies often underestimate how difficult it is to change employee routines. I’ve seen similar patterns with other technologies—without a clear incentive and user-friendly implementation, even powerful tools gather dust. Training and cultural shifts will be as crucial as the technology itself.

4. Speculative Investment Without Proven ROI (Sarah Myers West’s Concern)

I side with Sarah Myers West in seeing a speculative bubble forming around AI. The ‘FOMO’ (fear of missing out) driving Google, Meta, and OpenAI could mirror the dot-com bubble if these massive bets don’t yield practical returns. However, the competitive pressure may also accelerate innovation and force companies to find profitable applications faster.

My Take: A Killer App Will Emerge, But from Utility, Not Novelty

History shows that technological revolutions (electricity, the internet) often find their ‘killer apps’ years after their invention. AI’s breakthrough is likely to come from areas that blend human creativity with machine efficiency—perhaps in fields like personalized education, healthcare diagnostics, or autonomous operations. Right now, we are still in the ‘plumbing’ phase—building the infrastructure (large models, APIs, and integrations) that future killer apps will rely on.

In Summary:

• Short-term: Incremental improvements in productivity (e.g., Replit, Octopus Energy).

• Mid-term: Struggles with adoption and profitability (e.g., Synthesia’s challenges).

• Long-term: A disruptive use case will emerge, but probably from unexpected sectors.