- The Brainyacts
- Posts
- 025 | AI Speed Reader
025 | AI Speed Reader
Brainyacts #25
The Generative AI newsletter for legal pros everywhere.
š Hello! We have reached 730+ subscribers in 24 days! I cannot thank you enough for subscribing!! šš„
NEW SUBSCRIBERS, to read previous posts, go here.
My goal is to reach 1000 subscribers. Do you think I can make it? With your help, I can, so please share this newsletter with as many people as you can if you find it useful.
This is day 25 of 100 (our goal - 100 consecutive days!) and today we will:
hit on BloombergGPT
break down long-form content - Speed Read it via ChatGPT
share 1 tool and 1 event coming up
hear Kanye West makes an appearance
look at a GenAI Startup funding chart
Have big banks and financial services companies shot themselves in the foot by outsourcing financial data aggregation and analysis to the likes of Bloomberg?
As companies begin building their own LLMs (large language models), that generative AI can use to deliver insight, will we see more of them keep their data and build propriety LLMs? What is lost - what is gained?
We are truly at the frontier in this space.
The new BloombergGPT AI may be harbinger of the next wave of corporate AI. Current AIs are trained on web data (though firms can add their own training)
BloombergGPT is 52% either proprietary data or cleaned financial data. And it shows signs of being better at financial tasks.
ā Ethan Mollick (@emollick)
1:22 AM ā¢ Apr 1, 2023
Use Case: ChatGPTās character limit
ChatGPT has a character limit of roughly 2048. This is how much you can put in the prompt submission window. Now, this is not necessarily a hard and fast rule as the model runs on something called tokens.
The character limit and token limit are related, but the exact relationship depends on the language model and its configuration.
In general, a token refers to a single word or a subword unit in the input text, and the number of tokens is usually less than the number of characters due to tokenization. Therefore, a given character limit may correspond to a different number of tokens depending on the specific text and the language model's configuration.
But letās not get technical. A good rule of thumb is to not use more than 2000 words in one submission window that you want ChatGPT to digest.
If you go over this limit you will get the following error:
So what do you do if you want to āreadā or āunderstandā a large article or paper using only ChatGPT?
Itās fairly easy but requires a bit of setup.
ā¶ļøā¶ļøPROMPT
Step 1:
You need to do some prep work first. Find your long article. Copy-n-paste it into a Word or Google Doc. Find the word count.
You now want to break up the document into sections of roughly 2000 words. Err on the side of fewer. Make natural breaks - donāt cut off in the middle of a paragraph or sentence. Ideally, you can break it up into its own natural flow and sections. Just try to keep the flow.
Just put in a page break between each of the sections you made. This is just an easy reference for you to know where a section begins and ends.
Ok, now that you have that ready, letās move to the prompt.
ā¶ļøā¶ļøPROMPT
I will be submitting large portions of text. I want you to summarize by highlighting any frameworks, methodologies, and original ideas discussed. Further, I want you to extract the 3 most critical learning takeaways.
Here is the text: [paste it here]
This will return a nice summary of that section of text.
Do this for each section of text you have. IMPORTANT do this all in one chat session, do not create a new chat session each time.
Step 2
Once you have all of the text and the generated summaries in front of you, you have some options:
Summary of summaries: You can have ChatGPT recall each of the summaries it previously created and bring them together.
Executive summary: You can ask ChatGPT to create an executive summary based on the previous summaries it created.
To create a summary of summaries (thereby summarizing the whole document).
Copy each of ChatGPTās summaries into a new document.
Once you have that, use the following prompt and paste all these summaries after it.
ā¶ļøā¶ļøPROMPT
Please create a master summary of the following text. I want you to summarize by identifying all frameworks, methodologies, and original ideas discussed - use bullets for these. Further, I want you to extract the 3 most critical learning takeaways - number these 1,2 and 3.
[paste all of your summaries]
To create an executive summary You will prep as you did for a summary of summaries. Copy each of ChatGPTās summaries into a new document.
Once you have that, use the following prompt and paste all these summaries after it.
ā¶ļøā¶ļøPROMPT
Please create an executive summary of the following text. I want you to summarize by identifying all frameworks, methodologies, and original ideas discussed - use bullets for these. Further, I want you to extract the 3 most critical learning takeaways - number these 1,2 and 3.
[paste all of your summaries]
Here is a link to what ChatGPT generated for me. I used my Is Legal Design Bullsh*t? article.
Reviewing both the summary of summaries and the executive summary, I can say a reader would grasp key takeaways from either without reading this long article.
š£š„š¤Æ BOOM!!
Now you have an effective hack to digest long-form content. This is no replacement for actually reading every word but given that we swim in content all day, it is a helpful method.
AI Things
A tool:
BookAIChat: Use Chat to interact directly with books.
An event:
University of Colorado Law School: Generative AI and the Law - Exploring Generative AI and Law: ChatGPT, Midjourney, and Other Innovations
News you can use :
This is wild.
It shows you how fast this tech is moving and the crazy things it is and will be doing.
How will we be able to tell AI from real people anymore?
Watch and listen to this video.
This is the worse AI will ever be.
The Music industry is completely gonna be turned upside down.
Every artist will have their own custom trained model, and Spotify will own this space why?
š§µ A thread
ā Linus (āį“ā) (@LinusEkenstam)
10:27 AM ā¢ Mar 26, 2023
News you can lose: Generative AI companies raising $$$
Many will fail. Few will succeed. But that is the natural order of things. Hang on a bit and build your chops on the basic tools out there right now.
#GenerativeAI companies with >$5MM Raised (as on #Ma#March23) by Kelvin Mu #in#investorsOp#OpenAIp#OpenAIChatGPTh#ChatGPTda#datar#ArtificialIntelligenceu#automationi#digitalhealthe#healthtechoT#5Ge#technologyG#5Ge#telehealthh#mhealtha#databreachI#AIC#VCt#startupst#startupfundingtttwitter.com/i/web/status/1ā¦t
ā Dr. Monika Sonu I Founder Healthinnovationtoolbox (@sonu_monika)
7:51 AM ā¢ Mar 20, 2023
Was this newsletter useful? Help me to improve!With your feedback, I can improve the letter. Click on a link to vote: |
That's a wrap for today. Stay thirsty & see ya next time! If you want more, be sure to follow me on Twitter and LinkedIn.
DISCLAIMER: None of this is legal advice. This newsletter is strictly educational and is not legal advice or a solicitation to buy or sell any assets or to make any legal decisions. Please be careful and do your own research.