Five reasons why generative AI will outpace the growth of traditional data science
The overarching objective of generative AI (genAI)is simplification, removing the need for individuals to possess an extensiveskill set to implement its capabilities. Its applications span across a myriadof sectors, encompassing customer interactions, financial operations, salesstrategies, and more.
While many organizations are expected toleverage large language model (LLM) technologies and generative AI for diverseapplications, the distinction arises in the process of building LLMs. Unlikeconventional algorithms, LLMs possess broader applicability, making them apowerful tool that can address numerous use cases across diverse domains,further amplifying the pace of genAI's integration into various industries. Asa result, I expect we’ll see genAI far outpace the growth of data science for fivekey reasons.
1. GenAI is standing on the foundationsalready laid by AI
First and foremost, genAI stands to growfar more quickly because it’s building on groundwork that’s already been donewhen developing earlier technologies. While it would be immensely difficult tocreate a new technology capable of simultaneously analyzing data, makingpredictions, and generating new content of its own, because AI has been aroundfor years, this change in direction is merely an enhancement and extension of someexisting capabilities. Instead, we can consider genAI a subset of AI, which meansthat a lot of the trial, error, and initial work is complete, giving it asubstantial leg up in terms of its adoption.
2. GenAI benefits more users thantraditional AI
Second, the use case for genAI is much morewidespread than it was for traditional AI, which benefitted a smaller segmentof users. We can think of this as depth vs. breadth, where traditional AI cango deeper while genAI has broader applications. For instance, traditional AIwas typically deployed for specific use cases: medical research, deep analyticswork to gain insights, optimization of certain business processes. This left itwith a very small target audience, and most other users weren’t able to use AIfor their business needs. On the other hand, genAI can be put to work by reallyany user for a huge range of problems, both big and small. A data scientist canuse genAI just as easily as a business user—like a social media specialistbuilding out a content calendar, or a salesperson looking for help refiningtheir pitch. Add in natural language processing capabilities, we’ll see rapidscaling of genAI across users and use cases in the coming years.
3. GenAI is much easier to use thanks toits open-source nature
Third, at the end of the day, the use casesand how technology is exposed to the public make all the difference. And whiletraditional AI and ML models were still open source, adoption and using thosemodels required some strong expertise that generally only data scientists held.Whereas with genAI, we’re seeing much greater ease of use and integrations,including APIs that provide simple frameworks, so all you need to do is plugand play.
We can compare genAI to Apple’s App Store,which revolutionized what the developer community can build and deploy. BecauseApple provided the SDK, now developers have to worry less about going to marketand can instead focus on building and developing applications. It’s similarwith genAI; presented with an API, a model, and other ways people are using it,a user doesn’t need to have deep expertise in machine leaning. Instead, allthey need to know is how to use the API to build their own graph or applications.
4. The multimodality of LLMs means theysolve a greater number of user problems
Beyond these points, the multimodality ofLLMs means they can take in a range of inputs—including text, images, andaudio—and use them to solve a broader range of user problems. An LLM canproduce explanations, generate creative content, analyze visual data, and evenassist with tasks that require a combination of text and images, making ituseful for a programmer needing help with Python, a marketing professionallooking for ideas for an engaging email, or a data scientist conductingsentiment analysis. That wider array of applications, enabled by multimodality,means that we’re seeing much faster adoption of genAI as compared totraditional AI.
5. LLMs are pretrained with a diverseset of datasets
With traditional AI models, datapreparation was an enormous undertaking. Almost 60-70% of the work required wasspent on data engineering alone to prepare the data to feed and train themodel. With genAI, that issue is largely gone, because LLMs have pretrainedmodels. There’s no data to be fed, which means that a lot of the dataengineering pre-work has been cut down significantly. Not only does thisfacilitate faster adoption and easier deployment among trained data scientists,but it also helps nontechnical users hit the ground running in their user ofgenAI.
For these five reasons, genAI will rapidlyoutpace the growth of traditional data science. As it continues to evolve andexperience wider adoption rates, genAI will revolutionize business processes,allowing more users to take advantage of its unprecedented capabilities increativity, efficiency, and problem-solving.
