The future of generative AI is niche, not generalized

For example, AI algorithms can learn from web activity and user data to interpret customers’ opinions towards a company and its products or services. The computer-generated voice is helpful to develop video voiceovers, audible clips, and narrations for companies and individuals. AI is used in extraordinary ways to process low-resolution images and develop more precise, clearer, and detailed pictures. For example, Google published a blog post to let the world know they have created two models to turn low-resolution images into high-resolution images. While the most popular art NFTs are cartoons and memes, a new kind of NFT trend is emerging that leverages the power of AI and human imagination. Coined as AI-Generative Art, these non-fungible tokens use GANs to produce machine-based art images.

• And, while the technology offers tremendous promise, enterprises need to consider some of its challenges and limitations as they expand their use of the technology.
• Generative AI enables industries, including manufacturing, automotive, aerospace and defense, to design parts that are optimized to meet specific goals and constraints, such as performance, materials and manufacturing methods.
• The computer-generated voice is helpful to develop video voiceovers, audible clips, and narrations for companies and individuals.
• Eventually, AI-powered virtual assistants could become standard features in learning platforms by providing real-time support and feedback to learners as they progress through their courses.
• These technologies have potential to deliver transformational benefits over the next two to 10 years (see Figure 1).
• BCG and Google Cloud are excited about generative AI’s transformative capabilities, devoting significant resources to jointly help customers apply this breakthrough technology.

Roles in prompt engineering have recently emerged, as the need for that skill set rises alongside gen AI adoption, with 7 percent of respondents whose organizations have adopted AI reporting those hires in the past year. Lotis Blue Consulting’s Carroll believes generative AI will open numerous opportunities for fine-tuning domain-specific applications. For example, generative AI could extract insights from medical publications on a disease condition or automate mind-numbing query response typing work in customer service centers. LLMs could ingest industry-specific information to provide insight for domain-specific workflows. For IT decision-makers, the emphasis is moving from exploring the cool, new technology to identifying good data for training customers on LLMs for their apps without introducing operational or reputational risks to processes. “This may well be the catalyst that IT leaders needed to change the paradigm on data quality, making the business case for investing in building high-quality data assets,” Carroll said.

What are some practical uses of generative AI today?

It may still account for just under 31 percent of time spent, driven by growth in sectors such as transportation services, construction, and healthcare. One emerging application of LLMs is to employ them as a means of managing text-based (or potentially image or video-based) knowledge within an organization. The labor intensiveness genrative ai involved in creating structured knowledge bases has made large-scale knowledge management difficult for many large companies. However, some research has suggested that LLMs can be effective at managing an organization’s knowledge when model training is fine-tuned on a specific body of text-based knowledge within the organization.

Practically every enterprise app and service is adopting generative AI in some capacity today. And, while the technology offers tremendous promise, enterprises need to consider some of its challenges and limitations as they expand their use of the technology. Many of the first limitations slow down apps, while others might create real problems, like AI hallucinations, where generative AI apps make up content that’s not tied to facts. Recent examples of AI hallucinations are Google’s Bard incorrectly stating the James Webb Space Telescope took the first pictures of an exoplanet and the case of an Australian mayor suing OpenAI for defamation after ChatGPT said he had been jailed for bribery. Jonathan Watson, CTO at legal practice platform Clio, also attributes the explosion of generative AI to recent advances in generative models, such as generative adversarial networks and variational autoencoders, capable of generating high-quality outputs. In addition, generative AI has many applications, such as music, art, gaming and healthcare, that make it more attractive to the broader population.

Companies Need to Leverage Ecosystems to Deploy Generative AI

With these APIs, any application — from mobile apps to enterprise software — can use generative AI to enhance an application. Microsoft and Salesforce are already experimenting with new ways to infuse AI into productivity and CRM apps. When we had 40 of McKinsey’s own developers test generative AI–based tools, we found impressive speed gains for many common developer tasks. Documenting code functionality for maintainability (which considers how easily code can be improved) can be completed in half the time, writing new code in nearly half the time, and optimizing existing code (called code refactoring) in nearly two-thirds the time.

Founder of the DevEducation project

The ability to scale AI applications continues to challenge businesses across industries. Our collaboration with Intel brings together BCG’s transformation expertise, BCG X’s engineering capabilities, and Intel’s AI hardware and software in order to rapidly create genrative ai enterprise-grade generative AI solutions for our clients—securely and responsibly. Filling the jobs of the future is an opportunity to make the labor market more inclusive. Employers may need to reconsider whether some credential requirements are really necessary.

What are the best practices for using generative AI?

Since the release of ChatGPT in November 2022, it’s been all over the headlines, and businesses are racing to capture its value. Within the technology’s first few months, McKinsey research found that generative AI (gen AI) features stand to add up to $4.4 trillion to the global economy—annually. These technologies aid in providing valuable insights on the trends beyond conventional calculative analysis.

Given both industry and public concerns with privacy and data management, being cautious rather than being seduced by the marketing efforts of big tech is eminently sensible. While AI high performers are not immune to the challenges of capturing value from AI, the results suggest that the difficulties they face reflect their relative AI maturity, while others struggle with the more foundational, strategic elements of AI adoption. Respondents at AI high performers most often point to models and tools, such as monitoring model performance in production and retraining models as needed over time, as their top challenge. By comparison, other respondents cite strategy issues, such as setting a clearly defined AI vision that is linked with business value or finding sufficient resources. Improvements in generative AI technology could help firms find ways to harness imperfect data, while mitigating privacy concerns and regulations. AI models are already impressively capable when it comes to generating music and mimicking human voices.

And AI21 Labs’ own research pegs the expenses for training a text-generating model with 1.5 billion parameters (i.e. variables that the model uses to generate and analyze text) at as much as $1.6 million. For reference, the predecessor to AI21 Labs’ Jurassic-2 model, Jurassic-1, contained 178 billion parameters. It’ll certainly need the money, given the capital-intensive nature of developing large language models. If the practice of enhanced personalized experiences is applied broadly, then we run the risk to lose the shared experience of watching the same film, reading the same book, and consuming the same news. In that case, it will be easier to create politically divisive viral content, and significant volumes of mis/disinformation, as the average quality of content declines alongside the share of authentic human content.