AI is coming to the workplace - fear not.

A few thoughts about the potential impact of some of the new AI tools on businesses with @get_carbon as an example but applicable to other startups, particularly those with limited capital or resources.

The low hanging fruit

In our marketing department, we use graphic designers for images but #dallle can produce some stunning images. I wanted to push something out highlighting annoyed customers queuing for an ATM. These three images in a digital art style were done in less than 3 minutes.

At the extreme, we may no longer need graphic designers; more likely this will help graphic designers in being more efficient right now and helping in generating ideas. But its a game changer and allows startups with nothing to start quickly - and these are free for now

Branding and/or customer research are equally low hanging fruit; I asked #ChatGPT the question "what differentiates Carbon as a bank in Nigeria" and did the same for other competitors. Kuda, Fairmoney, Branch, Sparkle and for now it gave similar responses because of the limited training but singled out Carbon for its mobile-based credit card. The key point is that the extensive nature of the training data ultimately can help in research in learning about our brand positioning OR from a different perspective, provide insights as to what the market thinks

But its probably in customer service that we will see the biggest impact. The dirty truth about digital businesses is that there is significant manual/ analog components with large teams. Imagine training ChatGPT on your ideal answers for your most frequent customer problems. This is a chatbot on steroids

Looking Ahead

It gets more exciting though when we look at more complex activities like programming. In ChatGPT You could provide the following prompt: "write python code to calculate factorials" and you would get the following.

def factorial(n):
  if n == 0:
    return 1
  else:
    return n * factorial(n-1)

print(factorial(5))  # prints 120

or equally provide code and say "this code is not working like i expect — how do i fix it?" If the training data has been appropriate, you would likely get a viable solution. Just like in the marketing example this can help coders or replace coders in the future but this is just the beginning. Already @github claims that Copilot empowers coders to "complete tasks 55% faster by turning natural language prompts into coding suggestions"🤯. Imagine the opportunities especially with companies where coding talent is rare.

More exciting for me is in generating insights that are not obvious?

• Who are the most profitable customers?

• On what days to customers transact the most?

• Which state did we have the most losses in January and July?

Some of this natural language query already exists in different plaforms but the ability to query data independently, and at much lower cost will have a catalytic effect on productivity to a wide variety of companies

Conclusion

I need to stress that the potential impact listed above are surface level and there will be 2nd/3rd order effects. The key is to keep an open mind and be creative. Whilst these tools can replace humans, they are more likely to increase productivity. Exciting times indeed