Project management with ChatGPT: how neural networks will affect work and the market and how they can be used now
We figured out the technology and learned from experts whether projects and products should worry about neural networks.
Author: Nathanial Singh
02/03/23
Author: Nathanial Singh
02/03/23
Chatbot chatbot from Openair is frighteningly good: it can solve an equation, write an article, resume or even a diploma. We, the school for PMCLUB products and products, decided with a cold mind to see if a chatbot is really so scary or, conversely, it can simplify and speed up work on projects.
To do this, we talked to experts and tested for ourselves what ChatGPT is capable of.
In short: what is it about
ChatGPT is a "wrapper" of the GPT—3 language model, which can simulate a human conversation by learning from different sources. The technology was developed by Open AI. Microsoft has invested up to $10 billion in it to integrate the model's tools into its Bing search engine. According to rumors, "under the hood" he uses GPT-4 — a more advanced model trained on fresh data.
The "smart" assistant can answer any question on the Bing page — the user does not need to go to other sites. The main problem with this language model is that sometimes it gives "fictitious" erroneous answers to queries. Microsoft is currently working on fact-checking the bot.
Such systems cannot conduct a performance review, health check, or track productivity. These are too complex and complex tasks with a million parameters for which there are no single solutions.
At the same time, there are small tasks where it is already possible to use language models, including GPT. For example:
Answer simple team questions. There are no such products yet, but in theory, you can use a chatbot so that team members first ask clarifying questions to him, and not pull the manager every time. A similar model works in the usual chatbots of online store technical support.
Edit emails and adapt them to the audience. For example, you can "feed" a draft of a letter to a bot and ask it to be beautifully edited in the style of "British English" of the 2000s for foreign colleagues or clients. This will improve communication.
Translate speech in live mode. Also about communication. Neural networks can already translate a conversation live during conferences — at least in the form of subtitles. Soon they will be able to translate with their voice.
Generate content. ChatGPT can write a draft presentation or a post on a social network, and Midjourney can generate a picture on request.
To do this, we talked to experts and tested for ourselves what ChatGPT is capable of.
In short: what is it about
ChatGPT is a "wrapper" of the GPT—3 language model, which can simulate a human conversation by learning from different sources. The technology was developed by Open AI. Microsoft has invested up to $10 billion in it to integrate the model's tools into its Bing search engine. According to rumors, "under the hood" he uses GPT-4 — a more advanced model trained on fresh data.
The "smart" assistant can answer any question on the Bing page — the user does not need to go to other sites. The main problem with this language model is that sometimes it gives "fictitious" erroneous answers to queries. Microsoft is currently working on fact-checking the bot.
Such systems cannot conduct a performance review, health check, or track productivity. These are too complex and complex tasks with a million parameters for which there are no single solutions.
At the same time, there are small tasks where it is already possible to use language models, including GPT. For example:
Answer simple team questions. There are no such products yet, but in theory, you can use a chatbot so that team members first ask clarifying questions to him, and not pull the manager every time. A similar model works in the usual chatbots of online store technical support.
Edit emails and adapt them to the audience. For example, you can "feed" a draft of a letter to a bot and ask it to be beautifully edited in the style of "British English" of the 2000s for foreign colleagues or clients. This will improve communication.
Translate speech in live mode. Also about communication. Neural networks can already translate a conversation live during conferences — at least in the form of subtitles. Soon they will be able to translate with their voice.
Generate content. ChatGPT can write a draft presentation or a post on a social network, and Midjourney can generate a picture on request.
Take the quiz and find out the best destination for your vacation.
Fill out the form and we will contact you soon.
