$10 Million AI on a $2 Budget

The recent consumerization of artificial intelligence means that companies can unlock valuable new use cases, even if they lack pricey data scientists and high-performance infrastructure.

With salaries for data scientists approaching seven figures — to say nothing of the cost of high-performance computing equipment — most organizations cannot realistically build out an artificial intelligence effort from scratch. Yet, with robust, consumer-facing AI tools popping up over the past couple of years, companies can often replicate the results of a $10 million AI investment for the cost of a handful of API calls.

We’ve seen this story before, with the rise of the personal computer in the 1980s and 1990s. Processing power that previously would have required warehouses full of computing equipment suddenly fit on a desktop, boosting productivity for organizations and individual users at an unprecedented level.

Generative AI tools like ChatGPT, I believe, will lead to the next great leap in productivity. Here are six ways that companies can take advantage of it.

1. Sentiment analysis. Recently, I gave an assignment to students in my class on applied generative AI for digital transformation at MIT. I asked them to create an AI application that could analyze 10,000 comments on a YouTube video to determine which were angry, which were positive, and which demanded a response. This sort of thing would have required an expensive, purpose-built application only a few years ago.

However, my students were able to do it with programmatic prompts and APIs that connect to OpenAI. The most profligate student spent only around two dollars! This is potentially transformative for businesses. For instance, leaders at a regional restaurant chain no longer need to hire data scientists if they want to find patterns in their online reviews; they just need a few bucks.  

2. Repurposing internal content. Anyone who has used ChatGPT to try to create marketing materials can attest that the technology, while impressive, still has some limitations. (For instance, it tends to make things up.) But by ingesting their own internal data into AI tools, companies can generate accurate marketing content in seconds. Even something as simple as a product spec sheet with the dimensions of a chair can be repurposed into something that adds value to a company’s marketing efforts.

3. Customer service. Many businesses are already relying on automatic email responders and chatbots to improve the response times and capacity of their customer service teams. If this is done poorly (for instance, with simple auto-responses), it can turn off customers. But by leveraging information from past interactions and the preferences of returning customers, companies can actually use AI to provide a more tailored, personalized experience than people often receive when dealing with human customer service agents.

4. Employee training. Imagine a company with a large engineering department, and with a stack of dense, textbook-like tomes for engineers to consult when they have a question. An emerging pattern is to ingest all of this information and create a conversational model that engineers can consult when they have questions. Rather than manually searching through hundreds of pages, engineers can simply ask the model a question and get a precise answer in seconds. This not only improves employee satisfaction but also reduces downtime, increases efficiency, and perhaps even leads to a shorter time-to-market for critical projects.

5. Image Recognition for Quality Control. AI tools like ChatGPT are now able to ingest and quickly analyze photographs. For instance, you can upload a photograph of your refrigerator, and the tool will not only inventory the ingredients but will also give you a list of recipes you can make with what you have. Understanding images has surprisingly powerful applications. Historically, computer vision has been limited to controlled environments. Now, anyone with a phone has the potential to get assistance with understanding the physical world, from home repair to product inspections.

6. Human resources. A growing number of HR professionals are using publicly available AI tools to screen resumes and cover letters to help winnow down a large pool of job candidates. This not only can speed up the process but can also help bring data-driven insights to the hiring process. Now, it is important not to accidentally introduce bias into the process, but AI tools can actually help here, as well. Organizations can task an AI program with sniffing out biases or other problems with another AI program.

Over the past couple of years, AI has been democratized to an astonishing degree, and it’s easy to forget that we’re still at the very beginning of this new era. The most impactful commercial use cases of consumerized AI tools likely haven’t even been discovered yet. But companies that begin exploring low-cost, high-value applications will reap benefits today and position themselves to continue innovating as the future unfolds.

Upcoming Events


Related Articles