3 things I know about ChatGPT
3 things I know about ChatGPT
Will ChatGPT help businesses do more with less and keep a tighter management on their labour costs?
One thing about me is I don’t like talking too much about futuristic or unproven technology especially because it’s so easy to go down the rabbit hole and it’s distracting. A lot of “hypetech” has come and gone or caused harm in the process (just do a Google Trend search on NFTs for example or certain Crypto coins). I have been most wary of technology in search of a problem “Uber for x”, “AI x”, etc. Startups in themselves are experiments and 90% fail however with ChatGPT the use cases are real so I wanted to share 3 things I know about it.
On 4th December the internet started talking about it and everyone had become an expert the next day The NY Times posted an article about it.
ChatGPT can create content in seconds
It’s being used by creative, marketing and sales teams already: I know one under-resourced marketing team who has been told there is no budget for a copywriter using ChatGPT as an alternative to writing email subject lines and copy.
It will get answers wrong, (or not the answer you expected), but only at a fraction of the price. Currently, if someone hasn’t figured out an answer yet. All it can do is make a fuzzy guess as to what the right deterministic answer is as opposed to actually going to first principles and figuring it out. This basically means it will get answers wrong.
There are some interesting trends that anyone can predict and it’s very obvious they’ll improve the experience by making it unobtrusive when it makes a suggestion that isn’t correct. Naturally, the model has room for improvement and will be trained to improve general accuracy, match your tone of voice and tap into the datasets you allow it to read like your emails, business complaints, etc.
It feels like it’s in the nice phase where people feel like good answers are amazing but it’ll be interesting to see if we’ll ever get to that place where it’s so accurate we trust it to make decisions that could potentially be costly financially or to life but I definitely see it shining in places where there is no real harm in getting the answer wrong - kind of like Grammarly or transcription assistants now as you can edit the answer and that trains the model.
Just imagine those people who DM you on LinkedIn and you used to be able to tell that they copied and pasted it and didn’t even try to understand what you do. What does that look like in their hands now?
Photo credit @rugiyatou
Upcoming Opportunities & Events
Automate your customer service and business operations to cut costs - Barclays x CLS - Thursday 2nd February 2023. Link
Funding for businesses in the creative industries - closes 15th Feb. Link
Thoughts & News I wanted to share with you
Microsoft extend their partnership and the investment is reported to be $10Bn it has announced that ChatGPT will be available on their cloud platform Azure so developers can build on it. Link
P&G Beauty Acquires Textured Hair Brand Mielle Organics. Link
Replay this conversation hosted by Lendoe about the nuances of cashflow management for e-commerce and online businesses. Link