Simulmedians Utilize ChatGPT for Data Analysis
Open AI’s ChatGPT, released in November of 2022, has taken the world by storm. Questions concerning the ethics of AI have surfaced, as well as the prospect of replacing countless jobs with this new technology. In fact, according to a study conducted by economists from Oxford University, Artificial Intelligence could replace nearly 50% of jobs in the United States.
There is an upside to the innovation and ingenuity behind ChatGPT, though. The AI-based chatbot has the ability to exponentially increase efficiency, especially in the world of tech. As a company operating in the AdTech space, Simulmedians have already found ways to utilize ChatGPT to optimize for efficiency, boost productivity, and deliver industry-leading results to clients.
ChatGPT in the Industry
One of the truly mind-blowing aspects of Artificial Intelligence is that it is improving exponentially with each new query. According to Reuters, ChatGPT has already reached more than 100 million monthly active users since its release in November of last year — a new record. Although there have been issues with bandwidth, due to an excessive number of users overloading Open AI’s servers, ChatGPT’s user base will continue to grow, and so will its capabilities.
ChatGPT isn’t the only competitor in the space. This past Tuesday, Google unveiled its newest chatbot, Bard, to a less-than-promising start. After giving an incorrect answer in a promotion on Twitter, Bard’s mistake cost Google’s parent company, Alphabet, over $100 billion in market capitalization as shares plunged on the bad news. Even this small mistake denotes the current limitations of Artificial Intelligence.
Microsoft, which is a primary stakeholder in Open AI, has already found a use case for ChatGPT: Adding it to the search engine, Bing, in an effort to supercharge its capabilities. More than that, Microsoft is looking towards a future that will include ChatGPT’s functionality in Outlook, Office 365, and Microsoft Teams — software tools used by millions every day. These are just a few examples of how the leaders of Big Tech are hoping to leverage AI to drive their products, and their customers, further.
It’s safe to say that Artificial Intelligence won’t be taking over our lives tomorrow, but it will certainly pervade nearly every aspect of our day-to-day lives sooner than we might think. From coding to writing to Zoom meetings, AI chatbots like ChatGPT will be a mainstay, and its creators envision a world in which human efficiency will be dramatically increased by this burgeoning technology.
Putting ChatGPT to Work at Simulmedia
At Simulmedia, our Data Science team is working with large sets of data on a daily basis, building models to analyze and deliver performant advertising campaigns for clients. Analyzing large datasets is an area in which ChatGPT excels, as its computing power allows for incredibly fast results, as well as the automation of certain routine tasks.
Alex Papiu, Chief Scientist at Simulmedia, explains, “The underlying brain behind ChatGPT is a massive language model known as GPT-3.5. This language model is a probabilistic system that has been trained on a vast amount of text data. Its main objective during training is to predict the next word in a sentence, given the context of the previous words.”
In fact, Alex has already found uses for ChatGPT in his work at Simulmedia. “I mostly use ChatGPT to generate code for generic, exploratory data analysis. This usually involves getting data from SQL, aggregating it, and plotting the results. It’s also fairly easy to have ChatGPT define a simple Machine Learning setup in Python - although, for more complicated Machine Learning tasks, it tends to decrease in quality.”
Below is an example of how ChatGPT uses its predictive capabilities to generate code in Python based on prompts entered by Alex.
The Final Word
It’s indisputable that Open AI’s chatbot is a significant leap forward in the world of tech. The prowess of ChatGPT is even more impressive when considering the fact that it contains more than 175 billion parameters. ChatGPT is in a constant state of learning, too, as its language model continues to improve as more data is ingested and evaluated.
Despite its impressive capabilities, Open AI’s chatbot is not without limitations. Alex notes, “It's important to understand that the model is probabilistic and does not possess true comprehension of reality or truth. As a result, it may produce occasional outputs that sound plausible, but are incorrect.” In other words, ChatGPT won’t replace Software Engineers and Data Scientists in the near future; human discernment, namely domain expertise, is still necessary to determine the validity of the chatbot’s responses.
The future looks bright for Artificial Intelligence, and ChatGPT is just the beginning. As leading tech companies invest billions in the race to dominate the realm of AI, the underlying technology, and its ability to learn and progress, will continue its exponential growth. The future of technology is coming. Are you ready?