Report: Generative AI Will Transition Tasks Everywhere

2023 will be remembered as the year Generative AI became mainstream. Marked by the meteoric rise of programs such as ChatGPT, generative AI is increasingly becoming a daily ally in the lives of many, both in the workplace and in personal lives.

Market intelligence firm Omdia has released a report on generative AI titled, “The Generative AI Revolution: Understanding, Innovating and Capitalizing,” The report noted that while generative AI first appeared almost a decade ago, in 2014, this year marks the beginning of generative AI deployment at scale. The scaling up phase involves much experimentation and innovation, but also confusion, missteps and false starts.

According to Omdia, the generative AI boom has capitalized on a confluence of technical developments. Increased compute power and cloud capabilities, for instance, acted as necessary precursors for generative AI model training at scale. The demand for accelerated computing is likely only to increase as appetites grow for more and bigger training models. Omdia forecasts an expanding need for faster CPUs, more powerful GPUs, and extended cloud computing resources.

Driving Hardware Development

According to the report, the potential economic and sustainability costs of this need for increased compute power will likely drive a long-term shift toward alternative compute architectures, including model compression and additional reliance on edge computing.

The increased compute power will simultaneously act as both a driver for and is dependent on access to vast quantities of training data. The report said even greater quantities of training data will be required moving forward. This in turn is creating a market opportunity for synthetic data vendors who are using generative AI to generate additional data for training other generative AI models.

However, Omdia warned this data could potentially create a toxic data loop in which errors or biases from original sources are reproduced and thus reinforced in subsequent synthetic data. The report said one way to mitigate this issue is training via smaller, smarter, fit-for-purpose datasets, along with improved data curation overall.

Ease of Use the Key

According to the report, the most important driver in the generative AI boom has been the emergence of easy-to-use interfaces on top of models, personified by ChatGPT and Stable Diffusion. This has eliminated the need for users to have advanced technical or programming skills.  

Besides the easy-to-use natural language interface, ChatGPT and generative AI programs are desirable because of their ability to immediately benefit users by providing near real-time answers to queries. Omdia also noted that generative AI programs provide personalization at scale that makes it even more appealing to users. Outputs can be as unique as the user requests and/or designed to fit user preferences and profiles.

Who Can Benefit

The report noted several areas where generative AI can have a beneficial impact. For instance, customer experience management in the form of chatbots, digital engagement tools, and interactive self-help solutions are early use cases to make use of generative AI’s natural language interface. Generative AI’s ability to build on a response in ways that mirror human conversation can potentially give automated customer service interactions more flexibility.

The report also says generative AI can benefit marketing teams. Numerous companies are already rushing to explore how ChatGPT and GAI functionality can power copywriting, generate unique images and visual assets, develop quick product summaries, create personalized buying experiences, power translation, and more. Potential use cases within healthcare and the medical field are also beginning to emerge. These include summarizing patient records, examining scans for abnormalities, and assisting in medical research, according to the report.

Not surprisingly, software development could be impacted by generative AI’s ability to help write better code more quickly. Omdia noted that GitHub Copilot, released in June 2021, along with subsequent rivals, have shown the ability to return working source code after receiving natural language prompts. Code optimization, debugging, and documentation are all tasks generative AI could help streamline. Looking ahead, the report predicts generative AI-powered coding could democratize software development by offering non-technical users a path to bring new products to life.d

The report concluded that evidence remains inconclusive regarding just how safe GAI-created code is, with some reports showing greater vulnerabilities and some noting the opposite.

Risks, Potential Market Barriers

Counterbalancing generative AI’s promises are problems that are characteristic to AI in general. Omdia noted that any biases or inaccuracies in the underlying data can be replicated within the information that generative AI returns. This potential to unknowingly reinforce erroneous or potentially harmful data, moreover, is complicated by the lack of explainability within generative AI systems. The report noted it is currently impossible to trace which sources a generative AI system has used to inform its answers or content, and researchers are not entirely sure even how a generative AI system derives its conclusions.

As with other forms of AI, fraud and abuse can also be part of generative AI. The report noted there could also be sabotage attempts by hackers, rogue investors, activists, or competitors generating large quantity of fake but plausible high-quality social media posts, reviews, and videos that disparage a company, its products, or leadership, all of which could lead to reputational and material losses.

Not surprisingly, the issue of generative AI posing challenges to employment was mentioned. Generative AI’s ability to help with routine task automation, content creation, and customer service assistance have also sparked fears of worker displacement and unemployment. Moreover, job displacement from generative AI is likely to cut across both blue- and white-collar work.

Dos and Don’ts

The report gives several recommendations for users of generative AI. For one, it recommends users build an internal expertise and understanding of the technologies underlying generative AI and the ecosystem itself. Companies lacking the internal resources to build up this knowledge or unable to rely on outside partners should ask tough questions about what’s underneath the hood of any generative AI-enabled product and press for clear documentation and answers.

Establishing clear internal guidelines and guardrails is also key. Given the potential risks that accompany the space, the report said companies should at minimum ensure that generative AI polices explicitly map back to corporate privacy, security, and ethical guidelines. Establishing a clear set of rules governing use and development of generative AI, along with enforcement mechanisms, can help create company-wide alignment during a transition period.

The report also calls for users to closely monitor a rapidly changing landscape. This means not just watching the industry itself but also keeping a close eye on regulatory discussions and enforcement. Governments and regulators are moving quickly, and ignorance will not be an excuse for companies that fall on the wrong side of new rules and limitations to generative AI.

Omdia also suggests generative AI users start with low-risk, small-scale explorations. By starting small, either through internal development or with lower-risk vendor implementations, corporations can develop first-hand expertise and experience in this critical space while limiting potential fallout or monetary investment should anything go wrong or the market shift suddenly.

Spencer Chin is a Senior Editor for Design News covering the electronics beat. He has many years of experience covering developments in components, semiconductors, subsystems, power, and other facets of electronics from both a business/supply-chain and technology perspective. He can be reached at [email protected].