MSFT teases BingGPT🤖, G announces Bard AI⚔️ + ChatGPT gets MBA, lays down law📚

Welcome to our inaugural issue -- we're glad you're here 😊

Hi folks, here’s your ChatGPT news from the week along with a few of our thoughts to noodle on. Reply to this email with any questions or comments!

1. Is this Microsoft’s ChatGPT-powered Bing?

(3 min read) Microsoft is reportedly integrating OpenAI’s chatbot ChatGPT into its Bing search engine, with plans to launch by the end of March.

• new UI spotted by @Owen_Yin (Owen’s write-up of the experience)

• search bar will be a chatbox with 1,000 character limit

• Bing will be able to process complex tasks and cite references

Thoughts: MSFT may have accidentally(?) given a glimpse into its version of the future of search. Unlike big G, Microsoft is not as concerned with the “reputational risk” of ChatGPT-like products in search. Bing is ~9% of the market (compared to Google’s 84%), so not a whole lot to lose…. and if users do find the Bing+ChatGPT option compelling, Microsoft’s first-mover advantage may prove valuable.

2. Google announces Bard A.I. in response to ChatGPT

(3 min read) Google announced an AI chatbot called Bard that the company will begin rolling out in the coming weeks. Bard will compete directly with rival ChatGPT.

• Bard is powered by the company’s large language model LaMDA

• “Next week, we’ll be enlisting every Googler to help shape Bard and contribute through a special company-wide dogfood,” Pichai wrote in the email to employees that was viewed by CNBC (dogfood: internal testing)

Thoughts: Well… that was fast. ChatGPT launched just 2 months ago, which makes this the quickest Google product announcement in memory. Coupled with the company-wide dog-fooding effort, this looks like an “all hands on deck” moment for Google. We’ll have a lot more thoughts in the next issue on the upcoming competition between these tech giants.

3. ChatGPT Gets an MBA

(7 min read) The AI-powered chatbot did better than expected on a Wharton exam. That’s something to get excited about, says the professor behind the experiment.

• see paper summary, and the full paper here

Thoughts: In retrospect, one of the more obvious “uses” for AI is passing exams which are largely based on memorization. The biggest drawback listed in the study is ChatGPT’s math mistakes, which have largely been addressed in latest release. New challenges for educators, but also opportunity for productivity growth: students should “imagine the new rather than tweaking the old”.

We are not running out of work. Just look around. Our doctors and nurses are overworked, there is a mental health crisis, a climate crisis, our classrooms are overcrowded. What is shifting is the efficiency frontier. ChatGPT could help us gain in productivity. It’s for us as a society to figure out what to do with it.

Christian Terwiesch

4. Hustle Bros are Jumping on AI Bandwagon

(6 min read) The world of financial influencers promise viewers they can use ChatGPT to make big bucks with no effort. The schemes they suggest are dubious, but reveal how the AI chatbot might erode our online world.

One of the most popular TikTok videos explaining how to get rich with ChatGPT notes that viewers have a few years at most before the public become aware of the technology and should jump in before the rest of the world catches up. “For the small few who implement I’ll see you in first class,” says the video’s caption.

Thoughts: Light reading. Want/need another get-rich-quick scheme? Consult with these YouTube/TikTok prophets. Having said that, there may indeed be much long-term wealth created via ChatGPT and LLMs… but mainly for those who can harness and deliver the productivity gains such systems unlock.

This is the future many fear for the web: AI-generated junk suffocating online platforms like algal blooms that choke the life out of ponds. 

5. A Judge Just Used ChatGPT to Make a Court Decision

(3 min read) A judge in Colombia used ChatGPT to make a court ruling, in what is apparently the first time a legal decision has been made with the help of an AI text generator—or at least, the first time we know about it.

Thoughts: Law, in principle (not in practice) = laws + case history + logic. There is therefore a high potential for leveraging LLMs and AI in this arena, as well as a need for it. But the dangers of exacerbating built-in systemic inequalities and injustices are very real…

6. ChatGPT Has Turned NVIDIA’s AI GPUs Into Literal Gold

(3 min read) The rising popularity of ChatGPT, an AI-powered language tool, has turned NVIDIA GPUs into literal gold for the booming AI business. Company stock soars over 40% this month.

Thoughts: While the article is a bit of hyperbole (how does one prove the direct correlation to stock increase), it does highlight an inevitable increase in adjacent business models.

Bonus: ChatGPT’s creators can’t figure out why it won’t talk about Trump

(6 min read) Even ChatGPT’s creators can’t figure out why it won’t answer certain questions — including queries about former U.S. President Donald Trump, according to people who work at creator OpenAI.

Background reading: On the Unexplainability and Incomprehensibility of AI (19 min)

Human being are finite in our abilities. For example our short term memory is about 7 units on average. In contrast, an AI can remember billions of items and their capacity to do so grows exponentially, while never infinite in a true mathematical sense, machine capabilities can be considered such (infinite) in comparison to ours. This is true for memory, compute speed and communication abilities.

Shown impossibility results present a number of problems for AI Safety. Evaluation and debugging of intelligent systems becomes much harder if their decisions are unexplainable/incomprehensible. In particular, in case of AI failures accurate explanations are necessary to understand the problem and reduce likelihood of future accidents. If all we have is a “black box” it is impossible to understand causes of failure and improve system safety.

Stay tuned: Google Search and AI event on February 8th

This week, Google will be holding an event about how it’s “using the power of AI to reimagine how people search for, explore and interact with information, making it more natural and intuitive than ever before to find what you need,” according to an invite sent to The Verge. The 40-minute event will be streamed on YouTube on February 8th at 8:30AM ET.

…aaand not to be outdone, Microsoft just announced their own press event on Feb 7th — more to come!

---