How to use generative AI in business
“(ChatGPT) democratized AI in a manner not previously seen while becoming by far the fastest-growing app ever.” McKinsey & Co
I have previously discussed how ChatGPT, and other generative AI applications have begun revolutionising the AI consumer landscape. Through these AI models anyone, regardless of their knowledge and skill level, can directly interact with AI applications and experience how they work. While their entertainment value is high, models like these could also be used in business contexts to create value. The artificial intelligence industry is already a rapidly expanding one, with AI adoption and investments growing steadily in the last years.
Understanding Generative AI
In my mind, there are two reasons generative AI has been such a game changer in the AI landscape. The first reason is its generative capacity. Unlike previous machine learning models, which were primarily trained to observe and classify patterns, these new foundational models can create and continue these patterns. ChatGPT was trained on immense amounts of data, and from all that data it learned how human sentence structure works and is therefore able to re-create “new” sentences. While it is incredible to “talk” to ChatGPT, in a way it is not much more than incredibly sophisticated autofill.
Another thing that distinguishes generative AI models, is that they are generalists. Normally deep learning models are trained on a specific data set to perform strictly defined tasks. Foundational models, which form the basis for generative AI, are trained on large, unstructured amounts of data and can therefore be fine tuned into a wider variety of tasks.
It’s this generalism, built on top of a vast amount of data that allows ChatGPT to be used in so many different ways, and allows it to appear “creative” – being able to generate different answers to a prompt. This is also what has allowed for the democratisation of AI, as many have called it. Many machine learning models are employed in narrow contexts to fulfil the one task they are programmed for. But ChatGPT, like other generative AIs can be used by anyone, to answer any kind of prompt or question. This has allowed for far greater visibility and versatility in terms of AI usage, removing the skill-based barrier for entry.
Working with AI
One of the general fears around AI adoption, is its potential to replace human workers through automation. While there might be cases where this is appropriate, artificial intelligence should never be unsupervised in making critical decisions, especially when concerning human welfare. There is a growing conversation about responsible AI use, and how to recognise and mitigate the risks of using artificial intelligence.
Many of these risks originate within the core structure of a foundational model – which is trained on large, unmonitored volumes of data. With the usage of such data come concerns like privacy, intellectual property, and inherent biases, which can have implications for the outputs of such models. Text-based models like ChatGPT also tend to produce text that can range from situationally inappropriate to outright hallucinatory. Because of all these caveats, AI should never be entirely unsupervised and any critical content they produce should be proofread or double checked by an actual human.
Knowing this, it is easiest to integrate AI in ways that complement and enhance human capacities, rather than replacing them. The potential spectrum of use cases is vast: from improvements in research and healthcare delivery, to customisation of customer service and data analysis. For anyone interested in the industry capacities of artificial intelligence, as well as how to invest into AI, I recommend this McKinsey article.
Generative AI may also have implications for recruitment and hiring managers. Here too AI can function as a tool to automate some of the more tedious and time consuming tasks, such as writing job descriptions or sourcing candidates. Shifting responsibilities like these onto AI means that recruiters could dedicate more time to tasks which require a human touch, such as reaching out to these candidates.
Sources
Generative AI Use Cases for Industries and Enterprises
How Generative AI Can Help You Maximize Candidate Response In Hiring
LinkedIn: Generative AI May Play a Big Role in Recruiting
Responsible AI: Leading by Example | by BCG GAMMA editor | GAMMA — Part of BCG X | Medium
The 2023 Future of Recruiting report | LinkedIn Talent Solutions
The Rise of Generative AI in Talent Acquisition and Recruitment
The state of AI in 2022—and a half decade in review | McKinsey
What every CEO should know about generative AI | McKinsey
What is ChatGPT, DALL-E, and generative AI? | McKinsey
What is Generative AI and How Does it Impact Businesses? | BCG
How to use generative AI in business
“(ChatGPT) democratized AI in a manner not previously seen while becoming by far the fastest-growing app ever.” McKinsey & Co
I have previously discussed how ChatGPT, and other generative AI applications have begun revolutionising the AI consumer landscape. Through these AI models anyone, regardless of their knowledge and skill level, can directly interact with AI applications and experience how they work. While their entertainment value is high, models like these could also be used in business contexts to create value. The artificial intelligence industry is already a rapidly expanding one, with AI adoption and investments growing steadily in the last years.
Understanding Generative AI
In my mind, there are two reasons generative AI has been such a game changer in the AI landscape. The first reason is its generative capacity. Unlike previous machine learning models, which were primarily trained to observe and classify patterns, these new foundational models can create and continue these patterns. ChatGPT was trained on immense amounts of data, and from all that data it learned how human sentence structure works and is therefore able to re-create “new” sentences. While it is incredible to “talk” to ChatGPT, in a way it is not much more than incredibly sophisticated autofill.
Another thing that distinguishes generative AI models, is that they are generalists. Normally deep learning models are trained on a specific data set to perform strictly defined tasks. Foundational models, which form the basis for generative AI, are trained on large, unstructured amounts of data and can therefore be fine tuned into a wider variety of tasks.
It’s this generalism, built on top of a vast amount of data that allows ChatGPT to be used in so many different ways, and allows it to appear “creative” – being able to generate different answers to a prompt. This is also what has allowed for the democratisation of AI, as many have called it. Many machine learning models are employed in narrow contexts to fulfil the one task they are programmed for. But ChatGPT, like other generative AIs can be used by anyone, to answer any kind of prompt or question. This has allowed for far greater visibility and versatility in terms of AI usage, removing the skill-based barrier for entry.
Working with AI
One of the general fears around AI adoption, is its potential to replace human workers through automation. While there might be cases where this is appropriate, artificial intelligence should never be unsupervised in making critical decisions, especially when concerning human welfare. There is a growing conversation about responsible AI use, and how to recognise and mitigate the risks of using artificial intelligence.
Many of these risks originate within the core structure of a foundational model – which is trained on large, unmonitored volumes of data. With the usage of such data come concerns like privacy, intellectual property, and inherent biases, which can have implications for the outputs of such models. Text-based models like ChatGPT also tend to produce text that can range from situationally inappropriate to outright hallucinatory. Because of all these caveats, AI should never be entirely unsupervised and any critical content they produce should be proofread or double checked by an actual human.
Knowing this, it is easiest to integrate AI in ways that complement and enhance human capacities, rather than replacing them. The potential spectrum of use cases is vast: from improvements in research and healthcare delivery, to customisation of customer service and data analysis. For anyone interested in the industry capacities of artificial intelligence, as well as how to invest into AI, I recommend this McKinsey article.
Generative AI may also have implications for recruitment and hiring managers. Here too AI can function as a tool to automate some of the more tedious and time consuming tasks, such as writing job descriptions or sourcing candidates. Shifting responsibilities like these onto AI means that recruiters could dedicate more time to tasks which require a human touch, such as reaching out to these candidates.
Sources
Generative AI Use Cases for Industries and Enterprises
How Generative AI Can Help You Maximize Candidate Response In Hiring
LinkedIn: Generative AI May Play a Big Role in Recruiting
Responsible AI: Leading by Example | by BCG GAMMA editor | GAMMA — Part of BCG X | Medium
The 2023 Future of Recruiting report | LinkedIn Talent Solutions
The Rise of Generative AI in Talent Acquisition and Recruitment
The state of AI in 2022—and a half decade in review | McKinsey
What every CEO should know about generative AI | McKinsey
What is ChatGPT, DALL-E, and generative AI? | McKinsey
What is Generative AI and How Does it Impact Businesses? | BCG