---
title: "Goldman & MIT: AI Needs a Killer App to Prove It Isn’t a Bubble"
author: "Unique Research"
sourcePublication: "Unique Research (FFCAP English library)"
originalPublishedAt: "2024-07-12T16:15:00+08:00"
canonical: "https://ffcap.cn/en/research/20240712a064fa00"
source: "https://ffcap.cn/en/research/20240712a064fa00"
language: "en"
cnSourceUrl: "https://view.inews.qq.com/a/20240712A064FA00"
---

# Goldman & MIT: AI Needs a Killer App to Prove It Isn’t a Bubble

_Unique Research · Insight · July 12, 2024_

Is AI a fleeting tech bubble or the start of a lasting innovation wave? Opinions diverge. A Goldman Sachs and MIT discussion does not settle it: one Goldman economist and one MIT professor doubt transformative impact soon, while three other Goldman economists expect a “killer app” that further drives the industry. High development costs mean AI must pair with effective applications to justify investment and deliver long-term returns — a debate that spans traditional finance and crypto alike.

## Why transformation may be gradual

MIT’s **Daron Acemoglu** argues GenAI will keep advancing, but the path to transformative change will be more gradual and careful than many assume:

> Given the current focus and architecture of generative AI technology… truly transformative change will not happen quickly, and the likelihood of it happening within the next 10 years is extremely small (if it happens at all).
> 
> — Daron Acemoglu

Even if today’s LLMs reach “superintelligence,” he says, humans would still “remain in control,” because models lack many human capacities:

> It is hard to imagine an LLM having the same cognitive abilities as humans to pose questions, formulate solutions, then test those solutions and adapt them to new environments.
> 
> — Daron Acemoglu

A longer timeline, the piece notes, would help enterprises and other organizations take a more deliberate GenAI path — maximizing benefits while minimizing risk and disruption.

## Cost without the right applications

Goldman economist **Jim Covello** stresses that despite huge potential, AI’s expense means it must couple with applications that fully exploit its strengths. Without that — especially while AI still struggles with the complex challenges it faces — the case for the spend weakens:

> AI technology is very expensive. To justify those costs, the technology must be able to solve complex problems — which it was not designed to do.
> 
> — Jim Covello

## Beyond “picks and shovels”

Goldman economists **Joseph Briggs**, **Kash Rangan**, and **Eric Sheridan** are more optimistic: they believe AI can eventually “move beyond the current ‘picks and shovels’ stage and generate returns, even if AI’s ‘killer app’ has not yet appeared.” (The gold-rush metaphor: sellers of picks and shovels profited without digging — here, early AI is still mostly infra and platforms.)

Rangan grants that absolute dollar spend is high, but sees a more promising capex cycle than past ones:

> Today’s spending is indeed high in absolute (dollar) terms. But this capital-expenditure cycle appears more promising than previous ones.
> 
> — Kash Rangan

They do not name what the killer app might be, but say “there is still broad room for AI-related applications to develop,” and that AI will either “deliver on its promise” or fade — because, as they put it, “bubbles take a long time to burst.”

— END —

---

Editorial English edition published on UniqueCapital: https://ffcap.cn/en/research/20240712a064fa00
On-site reading page: https://ffcap.cn/en/research/20240712a064fa00