- Everyday AI
- Posts
- What AI experts are getting wrong (and right) about AI
What AI experts are getting wrong (and right) about AI
Everyday AI
June 27, 2024
š Subscribe Here | š£ Hire Us To Speak | š¤ Partner with Us | š¤ Grow with GenAI
Sup yāall! š
Kinda cool newsā¦.
I wrote a foreword for The Refractive ThinkerĀ® , an award-winning book series that features an anthology of doctoral scholars from various fields like business, education, healthcare, cybersecurity, social media, and marketing.
A few hours ago, the group released its 25th Anniversary edition, with this version focused on Artificial Intelligence.
I was honored to write the foreword, and the book right now is the #1 New Release for its category on Amazon.
Congrats to the team!
If you have $1.99, Iād say itās worth your money for the digital edition.
(Full disclosure, I get zero dollars and zero cents if you buy this. Just wanted to share helpful resources.)
āļø
Outsmart The Future
Today in Everyday AI
8 minute read
š Daily Podcast Episode: Lots of AI experts lead the charge in AI innovation. So what are they getting wrong and right? We break down our takeaways from recent conferences, events and discussions weāve participated in. Give it a listen.
šµļøāāļø Fresh Finds: Sam Altman gives update on GPT-5, Google translate adds 110 languages and Softbank forms an AI healthcare JV. Read on for Fresh Finds.
š Byte Sized Daily AI News: OpenAI releases CriticGPT and partners with TIME, U.S. ranked #1 in AI innovation, Amazon hits $2 trillion market cap and Goldman Sachs unveils AI coding tool. For that and more, read on for Byte Sized News.
š AI In 5: Make this ONE small change in how you access Microsoft Copilot and itāll make a world of a difference. See it here
š§ Learn & Leveraging AI: Weāve attended a few AI conferences and spoke to many AI experts. Hereās what theyāre all getting wrong and right. Keep reading for that!
ā©ļø Donāt miss out: Did you miss our last newsletter? We talked about ChatGPT app now out for all on Mac, Anthropic unveils Projects feature and 5 laws for AI success. Check it here!
What the AI experts are getting wrong (and right) about AI ā
What does the future of AI look like?
A lot of AI future casting starts as seedlings at AI conferences across the country.
Sometimes we're lucky enough to attend and speak at some of these conferences.
Sometimes, speakers and panelists are spot-on when talking about the future of AI.
And at times, they can miss the mark.
We're giving you all a quick recap of key takeaways of recent conferences, conversations and panels weāve participated in.
Join the conversation and ask Jordan questions on AI here.
Also on the pod today:
ā¢ Role and Importance of AI Experts š
ā¢ Trends in AI Startups and Public Companies š
ā¢ Ethical Considerations in AI Implementation āļø
Itāll be worth your 41 minutes:
Listen on our site:
Click to listen
Subscribe and listen on your favorite podcast platform
Listen on:
Upcoming Everyday AI Livestreams
Friday, June 27th at 7:30 am CST ā¬ļø
Hereās our favorite AI finds from across the web:
New AI Tool Spotlight ā Dappier is a marketplace for AI content and data rights, SEOrocket gives you SEO automation and Vmaker AI is an AI subtitle generator.
Trending in AI ā Sam Altman recently gave an update on GPT-5, saying the team wasnāt done training the model but itās apparently gonna be a looooot better than GPT-4.
Sam Altman says GPT-5 is not finished yet, but he expects it to be a "significant leap forward" and better at a wider variety of tasks than GPT-4
ā Tsarathustra (@tsarnick)
7:05 AM ā¢ Jun 27, 2024
Big Tech - Google Translate is adding support for 110 new languages.
AI partnerships ā Aside from OpenAI and Time magazine teaming up, Time also just announced a partnership with ElevenLabs.
AI in Healthcare ā Softbank is forming an AI healthcare joint venture with Tempus.
AI Startups ā Character.AI is now allowing users to talk with AI avatars over calls.
Google and AI ā Google released its new model, Gemma 2, to researchers and developers, as well as updated its Vertex AI platform.
Big Tech in AI ā According to a recent patent application, Google is trying to call dibs on Chains of Thought prompting. (Seriously)
Good catch here from @andrewwhite01
A patent on prompting methodology? Cool. Is Google gonna patent logical problem-solving next?
Guess we'll just call it: Logical Prompting Sequence or something. lolz.
ā Jordan Talks Everyday AI (@EverydayAI_)
4:45 PM ā¢ Jun 27, 2024
AI Coding and Development ā Langchain just announced LangGraph Cloud, their new infrastructure to help build more reliable AI agents at scale.
1. IMF Rates Countries on AI Adoption - Who's Leading the Pack? š
The International Monetary Fund (IMF) recently released a report assessing countries' readiness for artificial intelligence (AI) adoption. The US and Netherlands are at the forefront with a score of 77, while countries like South Sudan and Afghanistan are lagging behind with scores as low as 11.
India has emerged as a rising star in the AI landscape, surpassing the UK to become the fifth-largest economy by 2022.
2. Amazon Hits $2 Trillion Market Value Amid AI Optimism š
Amazon.com Inc has reached a significant milestone by hitting a market value of over $2 trillion. The e-commerce giant's stock rose 3.4% to $192.70, placing it in the elite company of tech giants like Microsoft, Apple, Nvidia, and Alphabet.
This achievement comes as U.S. stock indexes soar on the back of AI enthusiasm and anticipation of interest rate cuts. Amazon's strategic investments in AI technologies and cloud services position it well to capitalize on the evolving tech landscape.
3. Goldman Sachs Unveils AI Tool for Code Generation š§āš»ļø
Goldman Sachs is stepping up its tech game with the launch of a groundbreaking generative AI tool for code generation, part of their GS AI Platform. This innovative tool not only boosts developer efficiency by 20% but also ensures data safety and compliance with regulations, setting a new standard in the industry.
By forming strategic partnerships with tech giants like Microsoft and Google, Goldman Sachs is at the forefront of AI-driven development, reshaping the future of coding in the financial sector.
4. SoftBank's Vision Fund 2 Invests $3 Billion on Perplexity AI š°
SoftBank Group Corp's Vision Fund 2 is making waves with a whopping $3 billion investment in the AI startup Perplexity AI. Masayoshi Son's strategic move aims to elevate Perplexity AI's standing in the tech world and intensify competition with industry giants like Google.
This substantial investment heralds SoftBank's aggressive push into the realm of AI, signaling a high-stakes game in the ever-evolving tech landscape.
5. AI Answers Outsmart Professors in University Exams š§āš«ļø
A groundbreaking study at the University of Reading uncovered a sophisticated cheating scheme where AI-generated exam answers, submitted by fictitious student identities, outperformed real students without detection.
The research emphasized the growing capability of AI processors, like ChatGPT-4, to mimic human responses to a degree that poses significant challenges to the integrity of educational assessments.
6. OpenAI announces CriticGPT - A self-improving model? š±
OpenAI's latest innovation, CriticGPT, is shaking up the AI training world by boosting trainer performance and reducing errors in advanced AI systems.
Trained to critique ChatGPT responses, this tool outperforms human trainers in spotting mistakes and improving data quality for training models. Despite its limitations with short answers and complex errors, the integration of CriticGPT-like models signals a promising future for enhancing AI evaluation methods.
According to OpenAI's recent report, this development marks a significant step towards aligning AI systems with increasing complexity.
It appears OpenAI is testing CriticGPT internally and has not yet announced if this will be available to other companies, developers or model makers.
6. OpenAI announces partnership with Time
Time and OpenAI have just sealed a game-changing partnership, granting OpenAI access to a century's worth of Time's archives to enhance AI-generated responses.
This collaboration aims to make news content more accessible through AI tools while upholding journalistic integrity with proper source attribution. As news publishers increasingly team up with AI firms, the legal battle between media outlets like The New York Times and tech giants sets the stage for future AI and copyright infringement disputes.
The looming question for media companies now stands: resist AI or forge content partnerships to secure fair compensation for their work.
The ONE reason why you should be using Copilot in the Edge browser
ONE little tweak in you use Microsoft Copilot is gonna save you so much time and give you better results.
Isnāt that what youāre looking for here?
Check out today's AI in 5.
š¦¾How You Can Leverage:
AI experts are always right, right?
Naaaaah.
Hereās some real talk ā at Everyday AI, weāre lucky. Not only do we get to talk to some of the smartest people in AI (and bring you all the episodes) but we also get some inside scoops.
We get invited to speak at large conferences.
Chair panels.
Cover major tech events.
And share insights with other AI experts.
(Oh, if you want us to do any of those things, hereās the link.)
But it got us thinking ā a lot of the experts have it all wrong.
Weāve had some ongoing notes for MONTHS and decided it was time to dish and rant a bit.
Our founder, Jordan, was invited to host a panel at Chicago AI Week.
Full disclosure, our rant today is NOT directed at this one conference. (The conference was hosted by Xiaochen Zhang and 1871, which both did an amazing job.)
Instead, this is a collection of notes, observations and bones to pick that weāve compiled over the last few months from listening to āAI experts.ā
Ready for a recap on what AI experts are getting wrong (and right?)
And a lil ranting?
Letās do it. š
1 ā What AI experts are getting right ļøā
A lot, right?
Hereās the reality ā some leading AI experts have been toiling kinda silently in the dark for decades. And now that AI is front and center with the surgence of GenAI, our Artificial Intelligence experts have every mic and stage available to them.
Usually, itās a good thing. We think itās great.
Whether itās attending large AI gatherings or listening to a seasoned AI expert on a podcast, we get decades of insights jam-packed into one small package.
The skinny:
Hereās what we think the AI experts are getting right:
Safety ā Rightfully so, AI safety needs to be a high priority. Love the concept of needing more people to play defense.
Integration to AI first ā Thankfully, so much of the expert conversation has shifted from AI implementation to being AI first. The next step, which we should start talking about soon, is being AI native vs. AI first. In other words, not even having to make a āfirst stepā toward AI. Instead, starting there by default.
Specialties shining ā Experts with deep expertise in specific areas are finally getting the spotlight they deserve. Their ability to translate their expertise into leveraging generative AI is crucial.
2 ā What AI experts are getting wrong š
Also a lot. Lolz.
This partās problematic and plays off the upsides of whatās going right.
Just about anyone with solid experience around AI is being asked to talk about ALL things AI. Itās not always a good combination.
Those talking about LLM safety probably shouldnāt be talking about prompt engineering.
And AI ethicists probably shouldnāt be giving tactical advice on how to get the best results out of different GenAI platforms.
Yet, so many actually AI experts are being thrust into a variety of scenarios that they donāt necessarily have the expertise in.
(Side note: āI donāt knowā or āThatās not my expertiseā are amazing responses on panels or discussions.)
The skinny:
Hereās what we think experts are getting wrong, or just havenāt gotten right yet.
Flimsy Value Props: Many AI startups bet big on building niche large language models, but competing with industry giants like OpenAI might lead to a reality check. Also, donāt pitch that youāre building LLMs when if youāre tapping into an API or finetuning a model or adding RAG. Thatās not building a LLM.
Basic Gen AI Knowledge Gap: Month after month, this continues to shock me. Yes, thereās actual SUPER smart AI experts that donāt understand Generative AI at its core. (Seriously.)
Prompt Engineering Overlooked: Crafting effective prompts is crucial for maximizing large language model interactions, yet this aspect is often neglected in AI discussions. Instead, itās usually, āLook at this bad output I gotā or āListen to this anecdote about a prompt I ran once.ā Thatās a REALLY bad look.
Lack of Generalists: Another one that continues to shock. In the same way thereās Marketing or Advertising generalists, thereās not a lot of GenAI generalists. Just a bunch of specialists. (Hence, the point above about the basic GenAI knowledge gap.)
AI Content Detection Myth: Sorry yāall. If you or your company are pushing AI content detection, you lose all credibility. Instantly. LLM content cannot be accurately detected.
3 ā What we should ALL be focusing on š
Itās hard to skate to where the puck is gliding when no oneās ever played this game before.
But after countless hours of GenAI sessions, talks, and insights, hereās where we think companies and business leaders should be focusing on when it comes to GenAI:
The skinny:
Short sprints ā If your company is new to GenAI, donāt try and run year-long pilots. Start with short sprints with a measurable goal that are so easy to convey that an intern could run it. If you arenāt starting simple and tangible, it wonāt stick.
Ethics ā What happens WHEN GenAI is successful. What do you and your company do?
GenAI training ā All companies should be teaching the basics of GenAI and prompt engineering 101. (We do that here.)
Upskilling plan ā Studies show that up to 300 million jobs will be impacted by AI. If thatās the case, yes, there will be hundreds of millions of jobs created by AI. But weāre all kinda skipping the middle ground. What does upskilling look like? How can you keep and engage talented workers whose expertise will be diminished by AI?
ā
Numbers to watch
80%
Over 80% of businesses have embraced AI to some extent, viewing it as a core technology within their organizations.
Now This ā¦
Let us know your thoughts!
Vote to see live results
If our newsletter was sent at a more consistent time, would you read it more?Be honest. Our feelings won't be hurt |
Reply