Post by account_disabled on Sept 16, 2023 11:13:44 GMT
The 'deep love' of the media and general users for ChatGPT and other programs based on GPT-3 is unprecedented in IT history. Corporate IT decision-makers are also developing their own apps based on GPT-3.
ⓒ Getty Image Bank
So far so good. However, considering the Phone Number List Internet boom in the mid-1990s and the recent blockchain craze, it is true that there are concerns that companies may be hastily investing in things other than their strategic goals. In the early days of the World Wide Web, I once spoke with an executive about creating websites.
When asked why, for what purpose, and what he wanted to achieve, he did not give any specific purpose or goal. Instead, he gave a variety of ‘worst’ answers, including “One of the board members read the material and gave advice,” or “The CEO’s son keeps talking about the web,” or “Everyone else seems to be doing this.” It is quite similar to the situation regarding GPT-3 that we have recently heard about.
Strengths and limitations of ChatGPT from a strategic perspective
If you look at the history of IT, most of the technologies that are now widely known have ultimately proven to be strategically important. Not all, but most do. What about ChatGPT? First of all, it is clear that it is impressive. It is a huge database with an interface that plausibly imitates human communication, and can be called a kind of 'super-powerful intranet'. In fact, most of the information answered by ChatGPT cannot be found through a simple Google search. The key here is ‘search’. ChatGPT displays data that would take a person dozens of times to review Google search results all at once.
Another advantage, and one where much of the value can be found on the IT side, is human-mimicking interaction. This ultimately means that many coding projects do not require more basic programming tasks. Most programming projects begin with some line of business (LOB) executive or manager telling an IT expert, “Make the system do XYZ.”
What if CHatGPT took over some of this coding work and allowed you to generate code directly according to LOB guidelines without an IT expert? Some coding will require creativity and imagination and will continue to require a human touch, but (let's be honest) ChatGPT has the potential to take over the consuming and repetitive tasks that make up a significant portion of programming.
On the other hand, we also clearly identified the shortcomings of ChatGPT. These are absurd errors and completely false information released by the GPT-3 system. Until this issue is resolved, use of GPT-3 will be limited. Natural language interfaces are attractive, but for companies to rely on GPT-3 chat programs instead of humans to engage users in conversations about their products, they are inviting disaster.
ChatGPT as a coding interface
So how should you use ChatGPT? There are two ways to find the answer to this important question: prescriptive and open-ended. You can choose the method that is more suitable for you depending on your budget, business, and purpose.
First of all, the prescriptive approach is simple. Can provide more short-term results. For example, GPT-3 can help you find answers about what you can do now to help your business and launch some product or service that you have wanted for a long time but have not been able to realize. The open approach is more interesting. This is a way to creatively see what tasks GPT-3 can accomplish by giving it wide freedom to test GPT-3.
However, this method requires some limitations. Scott Castle, chief strategy officer at analytics company Sisense, said that if CIOs don't figure out what they want to do with ChatGPT, developers will end up focusing on the wrong ideas. “CIOs must filter strategically,” he said. “Otherwise, we are just mass producing Willy Wonka from the movie ‘Charlie and the Chocolate Factory,’” he said.
ⓒ Getty Image Bank
So far so good. However, considering the Phone Number List Internet boom in the mid-1990s and the recent blockchain craze, it is true that there are concerns that companies may be hastily investing in things other than their strategic goals. In the early days of the World Wide Web, I once spoke with an executive about creating websites.
When asked why, for what purpose, and what he wanted to achieve, he did not give any specific purpose or goal. Instead, he gave a variety of ‘worst’ answers, including “One of the board members read the material and gave advice,” or “The CEO’s son keeps talking about the web,” or “Everyone else seems to be doing this.” It is quite similar to the situation regarding GPT-3 that we have recently heard about.
Strengths and limitations of ChatGPT from a strategic perspective
If you look at the history of IT, most of the technologies that are now widely known have ultimately proven to be strategically important. Not all, but most do. What about ChatGPT? First of all, it is clear that it is impressive. It is a huge database with an interface that plausibly imitates human communication, and can be called a kind of 'super-powerful intranet'. In fact, most of the information answered by ChatGPT cannot be found through a simple Google search. The key here is ‘search’. ChatGPT displays data that would take a person dozens of times to review Google search results all at once.
Another advantage, and one where much of the value can be found on the IT side, is human-mimicking interaction. This ultimately means that many coding projects do not require more basic programming tasks. Most programming projects begin with some line of business (LOB) executive or manager telling an IT expert, “Make the system do XYZ.”
What if CHatGPT took over some of this coding work and allowed you to generate code directly according to LOB guidelines without an IT expert? Some coding will require creativity and imagination and will continue to require a human touch, but (let's be honest) ChatGPT has the potential to take over the consuming and repetitive tasks that make up a significant portion of programming.
On the other hand, we also clearly identified the shortcomings of ChatGPT. These are absurd errors and completely false information released by the GPT-3 system. Until this issue is resolved, use of GPT-3 will be limited. Natural language interfaces are attractive, but for companies to rely on GPT-3 chat programs instead of humans to engage users in conversations about their products, they are inviting disaster.
ChatGPT as a coding interface
So how should you use ChatGPT? There are two ways to find the answer to this important question: prescriptive and open-ended. You can choose the method that is more suitable for you depending on your budget, business, and purpose.
First of all, the prescriptive approach is simple. Can provide more short-term results. For example, GPT-3 can help you find answers about what you can do now to help your business and launch some product or service that you have wanted for a long time but have not been able to realize. The open approach is more interesting. This is a way to creatively see what tasks GPT-3 can accomplish by giving it wide freedom to test GPT-3.
However, this method requires some limitations. Scott Castle, chief strategy officer at analytics company Sisense, said that if CIOs don't figure out what they want to do with ChatGPT, developers will end up focusing on the wrong ideas. “CIOs must filter strategically,” he said. “Otherwise, we are just mass producing Willy Wonka from the movie ‘Charlie and the Chocolate Factory,’” he said.