Sam Altman (CEO of OpenAI) said, "AI will probably most likely end the world, but in the meantime, there will be great companies."
The rapidly changing technology landscape pressures management teams to transform organizations. Generative AI (GenAI), a cutting-edge technology, accelerates this need to change. According to an article in Forbes in January 2023, venture capital investments increased by a massive 425% in generative AI startups. ChatGPT acquired 100 million users in 2 months. The next closest TikTok dwarfs in comparison, which took 9 months. Democratization of AI at such a rapid scale makes one believe that this technology is here to stay and will be immensely disruptive. Furthermore, it's broad-reaching, helping stay-at-home parents as much as businesses become more efficient.
So, what is Generative AI?
Generative AI is a branch of artificial intelligence (deep learning in particular) that can create new content such as images, videos, text and music. GenAI has two types of models (1) GAN (Generative adversarial networks) and (2) GPT (Generative pre-trained transformers). The GANs are primarily used for deep fakes, and GPT is powering ChatGPT and is used to create content. GenAI is also used in fields like drug discovery, material science and robotics. It can generate art, solve complex problems and transform industries.
The evolution of GPT has been fascinating. It started in 2018 when OpenAI released GPT 1. The initial model had 117 Mn parameters and was trained on a dataset of approx. 8mn web pages. GPT2, launched in 2019, was trained on 1.5 Bn parameters and had a much larger dataset (40 Gb). In 2022 GPT-3 was released and trained on 570 Gb with 175 Bn parameters. In 2022 OpenAI also launched ChatGPT (GPT version 3.5). The data that is trained on is unknown. Shortly after GPT 3.5, GPT 4 was launched, vastly improving reasoning capabilities and greater accuracy in solving complex problems. Right now, ChatGPT has announced almost 85 plugins are growing every day. These plugins range from allowing users to use Instacart and link articles from the web where data is being pulled from to give more transparency.
The key features that cut across make for powerful opportunities.
While there are different applications for GenAI, many consistent commonalities make it a powerful tool for innovation and efficiency plays today and well into the future. The features that keep bubbling to the surface include:
Adaptability: Generative AI can adapt to new scenarios and data, making it more flexible and capable of handling complex tasks.
Creativity: It can generate new ideas, designs, and solutions, which can help organizations innovate and stay ahead of the competition.
Scalability: Generative AI can handle large datasets and tasks, making it suitable for enterprise-level applications.
Efficiency: Generative AI can quickly and accurately perform standard tasks, freeing employees to focus on higher-value tasks.
Automation: Generative AI can automate tasks across various business functions, reducing costs and improving efficiency.
Integration: It can be integrated with other technologies and systems, enhancing their capabilities and creating new opportunities.
Multi-Modality: It can handle different data types, including text, images, and audio, allowing for more comprehensive analysis and insights.
Depth: Generative AI can assist with analysis and inform decisions based on an extensive library of historical data, information sources and precedent
Why should the C-suite even care?
Since this technology is rapidly evolving, leadership must understand the business transformation and disruption GenAI can create. Industry leaders should consider opportunities to adopt and experiment with this technology at a pace while keeping risks in mind. In fact. The World Economic Forum says the C-suite needs to consider questions about AI from many angles, ranging from its strategic implications to new business risks.
The use case implementations are extensive; however, considering the end-to-end transformation would be the winning recipe. The essential questions for C-suite to ask are how is our AI strategy tied to our business and enterprise risk strategy? How are we quantifying value for our customers and stakeholders? What are the potential risks, and how would we mitigate those? Do we have the correct data? Do we have high-quality data? How are we going to use AI responsibly? Do we have the skills and talent? Do we have the proper governance? These questions and a well-thought-out strategy can create a point of future differentiation.
The benefits of Generative AI cannot be ignored.
Improved productivity and efficiency: Bloomberg says that Generative AI boosts worker productivity by 14% in the first real-world study. There are ways that GenAI can help workers analyze documents and get recommendations at their fingertips. This can help in improving productivity and efficiency.
Creative inspiration: Generative AI can act as a creative partner. It can help CEOs unlock untapped organizational opportunities, which AI couldn't help with historically. According to an article in the Economist, AI is already assisting artists in creating new poems, art and music.
Data augmentation: GenAI can produce synthetic data that closely resembles actual data. This is useful when there is insufficient training data to help other AI models with learning. An article in Nature argued that synthetic data could be better than real data.
Support services: With powerful NLP, GenAI uses large language models to build highly advanced chatbots. The chatbots can respond to queries in multiple languages instantly.
Serendipitous discoveries: GenAI can uncover unexpected patterns, relationships or insights that humans might not discover. In one example, MIT Technology Review highlights how AI is dreaming up drugs that no one has seen before.
The use cases of Generative AI are plentiful and growing.
While GPT is best for analyzing documents and generating content and language translation, below are some sample functional areas where GenAI can help businesses transform.
Human Resources: GenAI can be used across recruitment, onboarding, HR operations, exit and retirement. For example, AI can analyze online recruitment profiles and/or resumes and generate job descriptions. It can be used within HR operations to provide employee-specific policy information, analyze employee feedback, and develop personalized recommendations. For example, LinkedIn has used AI for years, matching job seekers with opportunities.
Legal: Within the legal function, companies use ChatGPT to write documents for M&A activities. A company called DoNotPay is using GenAI to help people contest parking tickets. It assists people in preparing court documents using GPT. We are also seeing courts use ChatGPT to get people answers to the questions they obtain from courts and complete legal forms.
Finance: The finance function within an organization can be transformed entirely across reporting, planning and analysis, treasury, accounting operations and tax. Within reporting, GenAI can be used to create narratives and footnote disclosures for 10k. It can not only help with trend analysis but also with anomalies in the patterns. It can be used in treasury for cash flow analysis, analyzing market data and managing investment portfolios.
Tax and Audit: The hot buttons are tax optimization by identifying tax deductions and credits, tax strategy by analyzing tax scenarios across different jurisdictions, tax compliance and automating tax-related processes to minimize risks. According to Ernst & Young, Real-time fraud detection is also a strong use of this technology. For audits, it can be used for regulatory and financial analysis.
What challenges are organizations faced with trying to embrace GenAI?
Proceed with caution because there are some challenges associated with AI, some pragmatic and some ethical. Leading experts debate how dangerous AI could be in the future, but there is no real consensus yet. However, there are a few dangers that experts agree upon.
Privacy: One of the biggest concerns ISACA experts cite is consumer data privacy, security, and AI. For example, Americans have a right to privacy, established in 1992 with the ratification of the International Covenant on Civil and Political Rights. But many companies already skirt data privacy violations with their collection and use practices. Experts worry this may increase as we start utilizing more AI.
Bias: It's a common myth that AI is inherently unbiased since it is a computer system. However, this is untrue. AI is only as unbiased as the data and people training the programs. So if the data is flawed, impartial, or biased, the resulting AI will also be biased.
Intellectual Property: There are many legal battles on copyright and IP. The top of mind is who owns the IP if the content is generated using GenAI. There are also legal cases, Writers Guild Strike, on ChatGPT as its training data is vastly available on the web, which is the creation of many writers and content creators.
Stand-up Costs: Developing and deploying AI systems can involve significant investments in hardware, software, data collection and processing can be barriers for some companies.
Regulatory Implications: Regulations could have an immense impact on adopting the technology. In a well-publicized case in Italy, they initially banned ChatGPT until OpenAI agreed to incorporate the country's requested changes.
Skills and Capabilities: The Financial Times claims that a global shortage of workers with skills and experience in areas such as deep learning, natural language processing and robotic process automation is slowing advances in AI. Companies need the skills and capabilities to use AI. Data scientists, ML engineers, and AI strategists are critical to a successful organization. According to Time, one of the hottest jobs in the market right now is a 'Prompt Engineer' who knows how to write prompts to get the correct answers from ChatGPT.
In summary, there is some hype but also a lot of reality. GenAI is here to stay and will only get better with future versions. Unlike previous technologies, AI can make increasingly complex decisions enabling new business opportunities. Still, AI decision-making comes with AI responsibility. Making responsible AI part of a business's operations requires adopting new practices and appropriate AI governance.
Authored by Dr. Lance Mortlock (EY Partner – Strategy & Adjunct Associate Professor), Samta Kapoor (EY Partner - Data Strategy & AI) and Pradeep Karpur (EY Partner – Data Analytics & AI).
Comments