• The Neuron
  • Posts
  • šŸ˜ŗAI expert predictions vs normie feels

šŸ˜ŗAI expert predictions vs normie feels

PLUS: The BEST AI prediction we've read...

Welcome, humans.

It turns out thereā€™s a pretty high chance that the US government just used AI to create its newly announced tariff policy:

The breakdown of ā€œreciprocalā€ tariffs on each country seemed to perplex most economistsā€”until James Surowiecki reverse-engineered the calculations, and then Rohit Krishnan found that all the major AI models landed on similar methods.

Translation? Somebody somewhere in the chain of command asked an AI to come up with the formula they ultimately used to dish out the tariffs.

No big dealā€¦ itā€™s not like the plan led to the worst single day stock market drop since Covid or anythingā€¦ but hey, somewhere in Silicon Valley, Jensen Huang is probably a little relievedā€”finally, a market sell off NVIDIA didn't cause!

Speaking of tariffs, those imposed on China, Taiwan, and South Korea could be about to hamper Big Tech's AI infrastructure investments by increasing data center costsā€”it looks like Microsoft has already delayed some new builds in Australia, the UK, and US.

Once the White House tells the AI what happened, their new policy will be like...

Hereā€™s what you need to know about AI today:

  • New research from Pew shows disconnect between experts and normies.

  • OpenAI made ChatGPT Plus free to US/Canadian students.

  • Spotify launched new GenAI ads.

  • DeepMind's Dreamer AI taught itself Minecraft.

AI experts and the public are miles apart on AI's futureā€”for better or worseā€¦

Plot twist: AI experts and the general public have wildly different takes on artificial intelligence. New Pew Research comparing 5.4K regular Americans with 1K AI experts just revealed a perception gap wider than the Mariana Trench.

The gap is striking: 56% of experts believe AI will positively impact the U.S. over the next 20 years versus only 17% of the public. While 47% of experts are more excited than concerned about AI in daily life, just 11% of Americans feel the same.

Most Americans (64%) believe AI will eliminate jobs, compared to only 39% of experts. In fact, 33% of experts expect no job changes, and 19% predict AI will create more jobs.

Both groups agree some jobs are at risk:

  • Cashiers (73% both predict fewer jobs).

  • Journalists (60% agreement).

  • Software engineers (50% agreement).

But they disagree on others:

  • Truck drivers: 62% of experts predict losses vs. 33% of public.

  • Teachers: 43% of public expects losses vs. 31% of experts.

  • Doctors: 28% of public predicts losses vs. 18% of experts.

There's also a representation problem. About 75% of experts believe men's views are well-represented in AI design versus 44% for women. For racial perspectives, 73% say White adults' views are well-accounted for, dropping to 50% for Asian adults, and only about a quarter for Black (27%) or Hispanic (25%) adults.

Interestingly, gender creates a massive outlook gap among the experts themselves: 63% of male experts predict positive societal impact versus only 36% of female experts.

One point of consensus? Regulation. Both experts (56%) and public (58%) worry more about insufficient government oversight than excessive regulation.

Our take: The most revealing finding is what's unsaidā€”the uncertainty regular people feel about AI. Between one-third and half of Americans answered ā€œnot sureā€ to many questions, indicating a knowledge gap the industry must address.

We wrote more about this on the websiteā€”check it out!

FROM OUR PARTNERS

This tech company grew 32,481%...

No, it's not Nvidiaā€¦ It's Mode Mobile, 2023ā€™s fastest-growing software company according to Deloitte.1

Their disruptive tech, the EarnPhone and EarnOS, have helped users earn and save an eye-popping $325M+, driving $60M+ in revenue and a massive 45M+ consumer base. And having secured partnerships with Walmart and Best Buy, Modeā€™s not stopping thereā€¦

Like Uber turned vehicles into income-generating assets, Mode is turning smartphones into an easy passive income source. The difference is that you have a chance to invest early in Modeā€™s pre-IPO offering3 at just $0.26/share.

Theyā€™ve just been granted the stock ticker $MODE by the Nasdaq2 and the time to invest at their current share price is running out.

Disclaimers

1 Mode Mobile recently received their ticker reservation with Nasdaq ($MODE), indicating an intent to IPO in the next 24 months. An intent to IPO is no guarantee that an actual IPO will occur.

2 The rankings are based on submitted applications and public company database research, with winners selected based on their fiscal-year revenue growth percentage over a three-year period.

3 A minimum investment of $1,950 is required to receive bonus shares. 100% bonus shares are offered on investments of $9,950+.

Prompt Tip of the Day

IDK about you, but weā€™re really impressed with Claude 3.7 (extended thinking version) and its ability to fact check itself (when asked). Try this prompt:

Please fact check each fact in the above output against the original sources to confirm they are accurate. Assume there are mistakes, so don't stop until you've checked every fact and found all mistakes.

Once you prompt it, click the little diagonal arrow next to the thinking dialogue box to expand its thoughts, and you can watch as it meticulously combs through each fact.

If it misses entire sections (a.k.a the lost in the middle problem), you can ask it to go back and check those sections specifically.

In our experience, as long as Claude has all of the original context in its context window, it rarely hallucinates. That said, Vectaraā€™s Hallucination Leaderboard still tracks Claudeā€™s hallucination rate at 4.5%.

The lowest hallucinator on the list? Shockingly, Gemini 2.0-Flash.

Treats To Try.

  1. Supaboard transforms your complex data into instant answers, actionable dashboards, and intelligent reports.

  2. Vapi creates voice AI agents that handle your phone calls and integrate with your existing toolsā€”free to try on the website (freaky).

  3. Tines helps you easily create no-code workflows that automate repetitive tasks across your organization's systems.

  4. OpenNutrition gives you instant access to accurate nutrition data through an open-source database for better meal tracking.

  5. Adaptive protects your organization from genAI social engineering attacks through realistic deepfake simulations across email, voice, and SMS channels (raised $43M from OpenAI).

  6. Vana lets you export your data from big tech platforms (here), contribute it to AI model training, and actually own a share of the models you help create.

Around the Horn.

  • OpenAI made ChatGPT Plus free for US and Canadian college students through May.

  • Amazonā€™s Kindle can now recap books in a series and its Shopping app has a new feature called Buy for Me completes purchases from other brand websites.

  • Runway raised $308M (and $536M+ total) to support more research on its video generator, hiring, and expansion of film production arm.

  • Spotify introduced new Gen AI Ads for creating scripts and voiceovers at no extra cost to advertisers.

  • DeepMind developed an AI system called Dreamer that figured out how to collect diamonds in Minecraft without being taught how to play, using reinforcement learning and a ā€œworld modelā€ to imagine future scenarios.

FROM OUR PARTNERS

Dell and NVIDIA explaining the value of local GPU compute across all industries  

Wonder why you should have your GPU compute locally vs. the cloud? Dell's new podcast ā€œReshaping Workflowsā€ shows exactly how professionals use Dell Pro Max workstations powered by NVIDIA RTX GPUs in real-world scenarios.

Intelligent Insights

  • If you watch one AI video this weekend, watch this one. Itā€™s 3 hours long (sorry!) but itā€™s probably the most impressive (and likely?) prediction of what comes next if AI research hits an ā€œintelligence explosionā€ (full prediction).

  • Intelligence analysts are increasingly over-relying on AI, causing a dangerous decline in critical thinking skills as investigators offload cognitive workā€”potentially undermining the entire field's integrity (cites this paper).

  • A team of law professors argued in favor or training AI on pirated content as ā€œfair useā€ (amicus brief).

  • This is crazyā€”video game cheats have apparently evolved from simple memory-reading ā€œaimbotsā€ to advanced AI-powered aim assistance that runs on separate hardware and are nearly undetectable by anti-cheat software.

  • Check out this great dialogue shared by ResObscura between a historian and an early AI scientist.

A Cat's Commentary.

Thatā€™s all for today, for more AI treats, check out our website.

The best way to support us is by checking out our sponsorsā€”todayā€™s are Mode Mobile and Dell.

What'd you think of today's email?

Login or Subscribe to participate in polls.