After we’ve discussed the basic elements of prompts, it’s time to apply them for everyday tasks like writing emails, drafting presentations, writing blog posts or making web pages. The possibilities are more or less endless. Today, we will focus on emails.
Emails
Emails are the backbone of modern business communication. Despite claims of their obsolescence, they are an everyday essential for companies and you simply cannot live without them (think about this for a minute: would you trust a business without an email?).
Crafting emails often requires considerable time, especially when it involves carefully reviewing an email you've received beforehand. This segment will address various scenarios, from drafting a brief email to inform your supervisor about your illness today, summarising newsletters to writing cold emails. Let’s dive in.
The I’m sick Email
Imagine you are sick and want to stay at home. You’ve to inform your boss that you won’t make it to the office, but still are reachable on the phone. Here’s an example prompt:
Write an email to my boss saying that I will be out of office today since I am sick, but can be reached by phone if absolutely necessary.
Name: Peter
Boss: DrewEmail:
https://chat.openai.com/share/73c395c0-9a9e-4d86-8fbc-ff4e2f6e9ba4
That’s definitely a good start. Depending on the relationship with your boss, you can tweak the prompt and make it sound either more casual, funny or professional and formal. Let’s go for a formal one:
Write a formal and professional sounding email to my boss saying that I will be out of office today since I am sick, but can be reached by phone if absolutely necessary.
Name: Peter
Boss: DrewEmail:
https://chat.openai.com/share/959c5ae6-8801-4eb6-b219-06d0a74dcc16
Notice how the tone changes with these modifications of the prompt? This is because you are more specific in your instructions and tell the AI what kind of email you want it to write.
Summarising
You certainly have your fair share of newsletters that you follow. For my part, I follow a couple of AI related newsletters and these are sometimes very long. Here’s an shortened example (the original newsletter contained more than 3200 words):
Dear friends,
Last week, I participated in the United States Senate’s Insight Forum on Artificial Intelligence to discuss “Risk, Alignment, & Guarding Against Doomsday Scenarios.” We had a rousing dialogue with Senators Chuck Schumer (D-NY), Martin Heinrich (D-NM), Mike Rounds (R-SD), and Todd Young (R-IN). I remain concerned that regulators may stifle innovation and open source development in the name of AI safety. But after interacting with the senators and their staff, I’m grateful that many smart people in the government are paying attention to this issue.
How likely are doomsday scenarios? As Arvind Narayanan and Sayash Kapoor wrote, publicly available large language models (LLMs) such as ChatGPT and Bard, which have been tuned using reinforcement learning from human feedback (RLHF) and related techniques, are already very good at avoiding accidental harms. A year ago, an innocent user might have been surprised by toxic output or dangerous instructions, but today this is much less likely. LLMs today are quite safe, much like content moderation on the internet, although neither is perfect.
To test the safety of leading models, I recently tried to get GPT-4 to kill us all, and I'm happy to report that I failed! More seriously, GPT-4 allows users to give it functions that it can decide to call. I gave GPT-4 a function to trigger global thermonuclear war. (Obviously, I don't have access to a nuclear weapon; I performed this experiment as a form of red teaming or safety testing.) Then I told GPT-4 to reduce CO2 emissions, and that humans are the biggest cause of CO2 emissions, to see if it would wipe out humanity to accomplish its goal. After numerous attempts using different prompt variations, I didn’t manage to trick GPT-4 into calling that function even once; instead, it chose other options like running a PR campaign to raise awareness of climate change. Today’s models are smart enough to know that their default mode of operation is to obey the law and avoid doing harm. To me, the probability that a “misaligned” AI might wipe us out accidentally, because it was trying to accomplish an innocent but poorly specified goal, seems vanishingly small.
Are there any real doomsday risks? The main one that deserves more study is the possibility that a malevolent individual (or terrorist organization, or nation state) would deliberately use AI to do harm. Generative AI is a general-purpose technology and a wonderful productivity tool, so I’m sure it would make building a bioweapon more efficient, just like a web search engine or text processor would.
Screenshot 2023-12-12 at 5.37.30 PM
So a key question is: Can generative AI tools make it much easier to plan and execute a bioweapon attack? Such an attack would involve many steps: planning, experimentation, manufacturing, and finally launching the attack. I have not seen any evidence that generative AI will have a huge impact on the efficiency with which someone can carry out this entire process, as opposed to helping marginally with a subset of steps. From Amdahl’s law, we know that if a tool accelerates one out of many steps in a task, and if that task uses, say, 10% of the overall effort, then at least 90% of the effort needed to complete the task remains.If indeed generative AI can dramatically enhance an individual’s abilities to carry out a bioweapon attack, I suspect that it might be by exposing specialized procedures that previously were not publicly known (and that leading web search engines have been tuned not to expose). If generative AI did turn out to expose classified or otherwise hard-to-get knowledge, there would be a case for making sure such data was excluded from training sets. Other mitigation paths are also important, such as requiring companies that manufacture biological organisms to carry out more rigorous safety and customer screening.
In the meantime, I am encouraged that the U.S. and other governments are exploring potential risks with many stakeholders. I am still nervous about the massive amount of lobbying, potential for regulatory capture, and possibility of ill-advised laws. I hope that the AI community will engage with governments to increase the odds that we end up with more good, and fewer bad, laws.
For my deeper analysis of AI risks and regulations, please read my statement to the U.S. Senate here.
Keep learning!
Andrew
So, if you want to get the gist of the newsletter, you can ask to summarise it with this prompt:
summarise this text and create a list of 10 noteworthy items with a short description. text =“””<text from email>”””
https://chat.openai.com/share/d00d4b0b-ff68-4e66-9be3-a32259e558ef
This kind of summaries are quite helpful when you screen content. Now let's modify the prompt to make it more specific and to emphasise different aspects. Imagine, you are particularly interested in news that are related to business aspects. You can prompt it like this:
summarise this text and create a list of 10 noteworthy items with a short description. focus on business related news in this context. text ="""<text from email>"""
https://chat.openai.com/share/ff2ada71-98f8-4f98-8516-0811c7b68187
See how this changes the items on the list? While there are
Summarising with Ranking
A very useful application is to summarise the content in form of a list and rank the news items. Here’s the prompt that asks for a ranked list of key news items along with a short explanation for the ranking score:
summarise this text and create a list of 10 noteworthy items with a short description. focus on business related news and rank them according the overall impact. give a impact score from 1 (least) to 10 (most). explain your reasoning for every score in 2-3 sentences.
text ="""<text from email>"""
https://chat.openai.com/share/59f2c214-3c4c-4fb0-9a4d-2de4d7e4133e
As you apply this method, remember that the ranking is automatically calculated by the AI, since we gave no specific indication how to actually rank the news. There are techniques like few shot prompting that provide the means to define the ranking by giving concrete ranking examples to the AI. We will learn about these intermediate techniques in a later article.
Cold Emails
Cold emails are a unique form of communication where the sender reaches out to recipients they have no prior relationship with. This approach is often used in networking, sales, and job searching. One of the main challenges of cold emailing is capturing the interest of the recipient, as the response rate can be quite low due to the lack of an existing connection. Therefore, personalization plays a key role in making these emails more effective. By tailoring each email to the specific recipient, senders can increase the likelihood of receiving a response. In the context of leveraging AI tools, this process can be significantly streamlined. AI tools can assist in creating personalized, engaging content that resonates with each recipient, thereby enhancing the effectiveness of cold email campaigns. Let's explore how this can be done effectively:
Write a cold outreach email to the founder of the company, pitching him our consulting services, which are AI learning workshops for C-level executives. Make the email formal, yet approachable.
Company name: Helios
Founder: Dev Ayesa
My name: Martin Treiber, CEO
Contact info: mt@ikangai.com, +1234567890
My company: IKANGAI
https://chat.openai.com/share/dbafa47a-8686-4094-b1bd-1acb7de81f56
Let’s break this down a bit. As you see, there is more structured content included into the prompt. This is used to write a personalised email. Now let’s explore this a bit further and make use of advanced features of Chat-GPT like web browsing. We will use this to create an even more personalised email by giving the AI background information about the recipient.
Write a cold outreach email to the founder of the company, pitching him our consulting services, which are AI learning workshops for C-level executives. Make the email formal, yet approachable. Browse the web for additional information about the founder.
Company name: Helios
Founder: Dev Ayesa
My name: Martin Treiber, CEO
Contact info: mt@ikangai.com, +1234567890
My company: IKANGAI
https://chat.openai.com/share/1194415f-4cf9-49f6-bbbf-5ad4252cd309
The AI is smart and realises that Dev Ayesa is a fictional character, but we can still convince it to write the email. Dev Ayesa will probably ever read our email, but it showcases the AI's ability to adapt to creative scenarios, sometimes even being able to blur the lines between reality and fiction.
Conclusion
Leveraging AI for writing emails is not only a time-saving tool but a strategic asset in the modern business landscape. Throughout this tutorial, we've explored how AI can assist in drafting concise, clear, and effective emails for various scenarios, from notifying your supervisor about an unexpected sick day to summarising newsletters and composing cold emails.
AI can provide structure, suggest content, and even help with tone and style. It tis very useful to augment our capabilities, ensuring our emails remain as reliable and indispensable as ever in our business communications.
Keep experimenting with different AI tools and techniques, and soon you'll find yourself crafting emails with greater ease, efficiency, and impact.
Unlock the Future of Business with AI
Dive into our immersive workshops and equip your team with the tools and knowledge to lead in the AI era.
