Personal computers, smartphones and cloud services have created an entire ecosystem that drives innovation.
According to Professor MIT's paper, generative AI is poised to pick up its mantle, with indications that it will serve as a foundation platform that enables a variety of software applications and services. and Like past market disruption technologies, generative AI is beginning to enjoy the benefits of enabling evolving ecosystems and tools and frameworks of the infrastructure layer, as well as the rapidly growing set of applications.
The pieces are still together, but their confluence sets stages for generating AI to become a powerful enablement technology, and software developers and technology producers still need to learn how to train the system and embed the generated AI in their applications.
“Generating AI is the next chapter: a modern innovation platform,” Kusumano said. “But we're still early on.”
The gusts of generative AI ecosystem activity have not been lost in industry watchers where emerging technologies are expected to have a drastic impact on the global economy. Goldman Sachs predicts generative AI could lead to a 7% increase in gross domestic product, with productivity growth increasing by 1.5 percentage points over the decade.
The Generating AI Ecosystem begins to evolve
Like previous innovation platforms, generative AI momentum is driven by a multi-layered ecosystem. Infrastructure providers Like nvidia, it creates graphics processing units and other hardware, but other hardware provides cloud services to run generic AI software. Basic model, Or large language models are produced by major technology providers such as Openai, Meta, Google, and small specialized companies. applicationwhich is now growing at a rate of hundreds and is being released continuously by small and well-known players. Both target specific industries and horizontal products that target a larger audience.
Researchers said past innovation platforms such as PC operating systems and the online transaction market also benefited from the participation of key players. Furthermore, it has gained traction through a phenomenon known as the “network effect” and is beginning to catalyze the generated AI. When more users attract third-party applications, network effects occur. This attracts more users and more applications, repeating the cycle and driving exponential growth.
“We are certain of one thing: now Researchers are time for corporate, university, government and technology experts to think carefully and think together. “We all need to better understand the costs, benefits, trade-offs and potential dangers of genai as a new application platform and transformational technology that are shaping the general future.”
AI Executive Academy
Directly on MIT Sloan
Sign up now
Issues on the horizon
As they move forward, organizations need to consider the following challenges identified by researchers:
Concentration of market power. The downside of strong network effects is the reduced competition. Software developers tend to combine the most popular or accessible LLMs to build applications. Furthermore, only a few companies have the money to develop and train basic models. These dynamics are likely to ensure continued domination of technology giants like Google, Meta, Partners Microsoft and Openai, researchers write. Large players can sweep startups and open source providers out of the generative AI ecosystem. “The dominant platform can curb innovation through predatory pricing, harm users and competitors, and generate lasting monopoly profits for a small number of companies,” the researchers write.
Several trends, such as the adoption of standard interfaces for popular LLMs, can help minimize developer lock-in. Additionally, “startups like China's Deepseek with large-scale open-source language models could change the balance of power from the major US high-tech companies and make generative AI software cheaper,” Cusumano said.
Data Ownership and Privacy. Data privacy, bias, and content ownership issues that emerged in previous innovation platforms also apply to generation AI. Continuing Litigation – For example, the New York Times' lawsuits for copyright infringement against Openai and Microsoft have not yet been resolved after using news articles to train ChatGpt and Microsoft Copilot.
Hallucinations and information accuracy. LLM responses may be irrelevant, effectively incorrect, or contain harmful content. Reliability can be improved through the use of external tools, and the accuracy tends to improve as LLM size increases. Nevertheless, this ongoing challenge requires robust mechanisms in place for developers and users to check for LLM output and fix errors. Some of the options researchers point out are to keep humans in loops throughout the process and invest in tools designed to detect fake text, audio, or video instances.
Regulation vs. self-regulation. The most complex and new technologies require a combination of self-regulation and government regulations. Researchers argue that combined government and private sector surveillance, especially as high-tech giants continue to dominate generative AI technology. “Checks and balances must be made to mitigate the risk of regulatory capture by funded LLM providers,” the researchers wrote.
Economic and social disruption. Generating AI can overturn existing hierarchies and corporate heights, especially high tech. At the same time, broader disruptions are likely to clash with white-collar jobs, and technology affects employment situations and the nature of the job. Many professions (teachers, journalists, lawyers, stock traders, computer programmers, corporate planners) may find their jobs are outdated, reinforced, or significantly changed.
Environmental impact. The computing resources required for LLM training and response are enormous. Work is underway to make LLMS and the resulting applications more energy efficient, but businesses and governments need to closely assess the trade-offs between energy use and profits of generated AI.
Unintended consequences
Researchers warn that there is no crystal ball to predict where generative AI technology will go. This new platform for application development creates opportunities for automation and brings efficiency to many jobs, but there is the real possibility of personnel reductions and significant changes in work processes. The possibility of inappropriately generating AI use in high stakes scenarios can also be catastrophic.
With all that in mind, it is important for organizations to better understand the costs, benefits, trade-offs and potential dangers of generating AI as a new application platform and achieve a better understanding of the already shaping a common future. “This technology will be everything,” Cusumano said. “It's extensive and we're going to get used to it.”
Read the paper: Generation AI as a platform for application development
