🤔 Wiser! #131: ChatGPT Code Interpreter | Anthropic's Claude | Deepfake Scams | Threads v Twitter
Two big AI announcements this were were Code Interpreter from OpenAI, and Claude 2 from Anthropic. I give you my twopenneth. Plus, my first thoughts on Threads.
What's in Wiser! this week?
There were four main stories that preoccupied my attention this week. Two in AI, one was a deepfake story and the other in social media. The big one for me was OpenAI’s latest feature, Code Interpreter. This feature is only available for the ChatGPT Plus paid service but it’s worth the admission price for this feature alone. I’m very impressed and you can expect a lot of excitement about this new toy.
The second was Claude 2. Having waited what seems like a million years to get access to Claude, the large language model from Anthropic, I finally got my hands on it and it was an anti-climax. There must be more to it, but I didn’t find it on first pass. Third was a perfect example of how deepfake technology is being used in scams, demonstrating the downside of the explosion in generative AI.
Finally, there’s Threads from Meta. It had just launched as I was putting last week’s Wiser! to bed. Now I’ve had to time to get online and, first impressions, very good. Here’s The Thing, Zuck has been really smart, he’s made it look like Twitter but with instant access to over a billion users. He has been rewarded with 100 million users in the first week! That has never been done before!
For the WiserPLUS! members, I’ve continued the theme of delving deeper into the different large language models. Last week I wrote the Bard v ChatGPT comparison article (thanks for your feedback, glad it was helpful). This week, I looked much closer at Pi, which happens to be my favourite of all the AI chatbots.
Next week’s WiserPLUS! article will be “Five Reasons Why Threads Will Beat Twitter.”
Finally, if you’ve not downloaded your free eBook yet, The Utility of Emerging Tech, you can get it here and it's free for everyone who reads Wiser!
ATB, Rick
P.S. REMEMBER: Insight and Information Gives You Leverage!
👍 “Great read, Rick. For me, very informative." - Efi Pylarinou, global influencer, fintech and disruptive technologies.
My Conversation With Pi, the Human AI from Inflection
Of all the large language foundational models I'm using at the moment, which is just about all of them, Pi is my favourite. Simply because the conversation and engagement with Pi is so humanesque, so positive, so helpful and so enlightening. Pi is not in the headlines, but it will be.
Here's The Thing: When you engage with the likes of Bard, Bing, ChatGPT or Claude, it "feels" like you're in a distant and disconnected online chat with a faceless banking agent. It's effective, but void of human emotion or connection. It's not like your free-rolling with colleagues in the office or bantering with mates in the pub. The chat is a functional, master/slave, conversation. I no longer use please and polite conversation with ChatGPT, the machine responds just as well to an order as it does to a polite request.
But Pi is different. And it never fails to surprise me at how perfectly human it can be in its use of language. Always polite and respectful, Pi also has a sense of humour, a few ounces of humility and an enduring habit of being super positive about whatever task you've asked it to solve for you.
Anthropic release Claude 2: IMHO, disappointing!
I’ve been on the waitlist for Anthropic for months and finally I got access this week to the latest AI chatbot. Anthropic is a foundational AI “safety startup” based in San Francisco with a large language model called Claude. Unlike all the other AI system, Claude is not an acronym, simply the name they gave the AI. Anthropic have raised around $1.5 billion so far to build an AI that is capable of longer and safer conversations with humans.
Here’s the thing: I was disappointed with Claude! Claude is meant to be very capable, scoring 75% in the Bar Exam and coming above the 90th percentile in reading and writing college entrance exams. But it couldn’t write 10 sentences that ended in “apple”!
That, in itself, is not exceptional, I haven’t found any AI that can do it. The difference, IMHO, between Claude and all the other foundational LLMs, is that Claude didn’t appear to be able to do much at all, and certainly nothing new or different to the other main foundational large language models. But, it’s early days and I’ll keep looking for the purpose of Claude.
More on Claude: Scale, Arstechnica, New Vision
ChatGPT release game changing feature called Code Interpreter
This is exciting! OpenAI have opened up their Code Interpreter feature for all ChatGPT Plus users (which costs $20+tax a month). Code Interpreter gives ChatGPT the ability to run computer code, use files you upload to produce outputs, analyse data in any number of formats, even unstructured. It even has facial recognition. Code Interpreter eats your data and spits out analysis, insights, charts and visualisations and can even perform sophisticated math (something that large language models are not very good at.)
Here’s The Thing: ChatGPT has put the ability to do all sorts of data analysis and code-dependent tasks it couldn’t do before into the hands of everyday users. A set of capabilities way beyond the reach of the majority of humans! If you’re a small business, a sales rep, basically anyone that works with data and is trying to figure out where to focus their time, Code Interpreter is like the magic calculator for data analysis.
I’m really excited about it and wrote an article that you can read here for free.
Threads tops 100 million users as Twitter usage tanks
This time last week I wasn’t on Threads, but thanks to my VPN and Instagram account, I was up and running early Saturday morning in a jiffy. The master stroke was tying Threads to Instagram because it means that a user takes their social graph with them when they sign up to Threads. On day one they have all the same friends, contacts, history, preferences. It’s like being born a teenager and skipping the first 13 years of development.
My first impression is that I like it, which is something I never thought I'd say about a Facebook product. It’s unpolluted and clean. But it is also early days and missing a ton of features, the main one being the ability to search for trending hashtags and stories. That will come I’m sure. The point is that Mark Zuckerberg did the smartest thing when he built Threads. He made it look just like Twitter. No pretense, no fancy gimmicks, simply cut and paste someone else's good idea.
And suddenly, the Zuck is cool again, thanks to the single handed efforts of his arch-rival, Elon Musk.
Here’s The Thing: Musk can huff and puff all he likes about suing Zuckerberg for stealing Twitter’s IP, but the truth of the matter is that Musk has been undermining Twitter’s defences since he took over. We talk about building moats as the analogy for fending off competitive threats. Well, Musk has filled his moats, lowered the drawbridge and fired 80% of his foot solders, leaving fortress Twitter defenceless.
You can see it in the numbers. Twitter usage is down 11% YoY. The CEO of Cloudflare, the Internet company that helps deliver webpages faster, posted a screenshot that showed Internet activity levels for Twitter. They only go one-way, down 5% in the first 2 days post Threads! 📉
Side Note: Twitter rival Bluesky hits its first million installs. Bluesky, another Twitter rival, has hit 1 million subscribers. The only way to join Bluesky is by joining the waiting list (which I have, still waiting) or getting an existing user to give you an invite (I don’t know anybody on BlueSky!)
Here’s The Thing: launching a new social media platform is really tough. With all of Jack Dorsey’s personal credibility, significant wealth and the resources at his disposal, he hasn’t been able to do in 3 years what Zuckerberg has done in 3 months. Having said that, I’m optimistic about BlueSky, from what I’ve heard about it.
5 Reasons Why Threads Will Beat TwitterX
Martin Lewis and the deepfake video scam
A downside of the mainstream adoption of generative AI is the rapid rise in misinformation, disinformation and deepfake videos. In this video clip taken from UK morning TV, Martín Lewis talks about a deepfake scam advert made in his name, image and voice. As you can see in the video it’s convincingly persuasive that it’s Lewis doing the actual talking. But it’s all fake.
Here’s the thing: Martin Lewis is a highly trusted consumer champion in the UK. Anyone watching this thinking it’s Lewis would be taken in and be vulnerable to the scammers that made it. The point is that you don’t need to be Disney to make these fakes. I just went to see the latest Indiana Jones and the opening scenes where Harrison Ford looks 40 are incredible, it’s hard to see the join (and I was looking!) but that’s Disney!
In the case of the Lewis deepfake, these can be made for very little money given the number of generative AI platforms now on the market, all competing for a modest monthly subscription. By comparison, Disney spent $300 million making the latest Indiana Jones movie.
Some other news
Alibaba is pledging 5 day global delivery through Cainiao.
Lawmakers in Massachusetts are considering a ban on buying and selling location data. The bill, called the Location Shield Act, would be the first of its kind in the US. About time!
Elon Musk has launched an artificial intelligence startup, called xA, to “understand reality”.
Google's new terms of service claims the right to scape anything off the Internet to train its AI models. They've been doing it anyway, but this is heading off possible litigation (if it works!)
Google is developing an AI program to answer medical questions, competing with Microsoft to create widely-used products for clinicians.
The latest NFT trend is taking out loans on Rolexes. NFT lending protocol Arcade and escrow service 4K aim to create a global pawn shop of strangers, starting with Rolex and Patek Philippe watches.
Earlier this week, a US court ruled that Microsoft can go ahead with its $68.7 billion Activision Blizzard acquisition. The FTC are now appealing that ruling, although Microsoft hope to complete the deal by July 18th or face a renegotiation option from Activision.
Comedian Sarah Silverman and two other book authors accused OpenAI and Meta of copyright infringement by taking their content as training data for AI in lawsuits against the two companies.
Shopify CEO Tobi Lütke tweeted Shopify's AI vision and new commerce-specific AI digital assistant, Sidekick, which can chat with customers, make recommendations and encourage more shopping!
Chart of the Week
Activision Blizzard is a massive business and it’s no wonder that Microsoft has been patient and persistent in its pursuit of the gaming giant. The headline game may be Call of Duty, but its Candy Crush that’s the real gem. It’s been downloaded almost 3 billion times and is the 6th highest grossing mobile game of all time.
For more data driven insights, sign up for your own copy of Chartr.
Recommended Reading
Burn the Boats: Matt Higgins (a LinkedIn connection of mine), sent me his new book, Burn The Boats. Matt talks about my time on Shark Tank and how the burn the boats mindset took him from a 16 year old high school drop out in abject poverty to the set of the top business show in the world. Key takeaway: humans perform at our best when we give ourselves no option but Plan A. 👉 Read It Here.
AI Tool Report: Learn AI in 5 minutes a day. We'll teach you how to save time and earn more with AI. Join 55,000+ free daily readers from Tesla, Apple, A16z, Meta, & more. 👉 Sign Up Here.
Bitcoin Breakdown: A twice-weekly Bitcoin-focused newsletter with the highest signal-to-noise ratio in the space, carefully curated by an alien from the future. 👉 Sign Up Here.
TheFutureParty: We're a media brand creating content for creative professionals at the intersection of business and culture. 👉 Sign Up Here.
The utility of emerging technologies
This week's top stories about the utility of emerging technologies:
⛑️ The World Economic Forum List Metaverse For Mental Health In Top 10 Tech Trends. Read more on LinkedIn or Instagram
💋 Sephora Opens It's First Store Of The Future In Shanghai. Read more on LinkedIn or Instagram
🧠 Researchers Develop AI Tool To Test For Dementia. Read more on LinkedIn, or Instagram
👖 Fashion Consumer Brand J.Crew's Virtual Reality Store Experience. Read more on LinkedIn or Instagram
♾️To find out more about the use-cases of over 250 consumer brands in emerging technologies, read this...
Be Wiser! than your competition
Wiser! cuts through the noise and tells you what you need to know about tech and business. From disruptive technologies to emerging trends, from BigTech to social media, Wiser! makes sense of it all in plain and simple language so that you can be Wiser! than your competition and colleagues.
☕️ If you'd like to acknowledge my work without buying a subscription, you can BuyMeACoffee. Virtual coffees only cost €2. 🙏