AI Minds #7: 2023 State of AI Report

A review of Stanford's 2023 AI Index and State of AI Report.

Welcome (back) to AI Minds, a newsletter about the brainy and sometimes zany world of AI, brought to you by the Deepgram editorial team.

In each edition, we’ll highlight a special topic from the rapidly-evolving field of Language AI. We’ll also touch on timely issues the AI community is buzzing about. As a heads-up, you might want to brace yourself for the occasional spicy take or two along the way. 🌶️

Thanks for letting us crash your inbox; let’s party. 🎉

🌍 The Status Quo of Tech

The AI world released a few “State of…”  reports recently, and the whole world is watching. To catch up on the status quo would require reading hundreds upon hundreds of pages.

So, to save you some time, this edition of AI Minds compiles key takeaways from these major reports. However, we do highly recommend reading the original text when you get the chance! Here they are:

Stanford’s 2023 AI Index - The AI Index is an independent initiative at the Stanford Institute for Human-Centered Artificial Intelligence (HAI). . . .  The annual report tracks, collates, distills, and visualizes data relating to artificial intelligence.”

State of AI Report 2023 - The State of AI Report has released its sixth annual edition, covering five key “dimensions” of AI: Research, Industry, Policy, Safety, and Predictions. The Safety section is new this year.

State of Voice Technology in 2023 - With both voice- and audio-oriented AI on the rise, this report covers the value of voice tech, the common barriers/motivations for voice tech adoptions, and the intersection of LLMs with voice. The exact ROI on voice tech implementation and adoption may surprise you!

🐦️ Tweet Alert: State of AI Thread

Here’s a peek at what Twitter’s saying about the State of AI (Source: Andrej Karpathy)

📝 Key Takeaway #1: AI does the dirty work

Anyone who’s been perusing LinkedIn for a dream job has probably noticed something new. As the key takeaway from the Stanford AI index stated “The Demand for AI-related professional skills is increasing across virtually every Sector”.

Source: Stanford HAI (AI Index, Chapter 4)

🚛 What does the future of hiring look like in practice? 

Well, as the number of AI-related job postings has increased on average from 1.7% in 2021 to 1.9% in 2022, employers in the United States are increasingly looking for workers with AI-related skills.

ChatGPT and Github Co-pilot have become this generation’s gateway drugs technology to AI. (Check out our thoughts on copilots here). After tasting a nice productivity bump, early adopters are diving into the wide world of AI tooling.

Job loss concerns: Fact or Fiction?

And as for whether the general public should worry about being replaced by AI, the State of AI Report reveals that 27% of jobs are at “high-risk” of being completely substituted by AI. (See slide 137 here)

Meanwhile, policy makers are waiting for the complete arc of AI to unfold before stepping in to take action.

Certain stand-out voices in the industry—from Sam Altman to Mustafa Suleyman—have all therefore expressed support for a Universal Basic Income.

Source: State of AI Report, Slide 137

🤖 Attaining an AI skill set

Whether it’s a film director leveraging storyboarding tools like sotria, or a marketer who wants to set up a no-code AI-powered workflow with AirOps, the world of work will never be the same. 

Looks like creating AI selfies of your pet ferret might actually pay dividends after all. And if you’d like some tips and tricks in using AI to enhance your work, check out this video!

🌲 Key Takeaway #2: AI helps & harms environment

Source: Stanford HAI (AI Index, Chapter 2)

It’s no secret that technology directly impacts the environment. However, the extent to which the environment is affected depends on which technology we’re analyzing. When looking at AI specifically, new research reveals that, for example, “BLOOM’s training run emitted 25 times more carbon than a single air traveler on a one-way trip from New York to San Francisco.” (Read: a one-way trip from Wall Street to Silicon Valley).

In fact, we’ve analyzed the exact give-and-take of Carbon and AI in one of our latest articles here. In it, our very own Tife Sanusi notes: “Google, Microsoft, and OpenAI have all made promises to improve energy efficiency or pursue net zero emissions.”

🐦️ Tweet Alert: State of AI doesn’t touch on environment

Also, note that the State of AI Report 2023 does not cover the environment whatsoever. None of the five major sections of the report discuss carbon emissions or energy consumption. Perhaps we can hope to find more info in next year’s edition!

Note: We’ve also covered the environmental cost of AI in the previous edition of AI Minds!

💸 Key Takeaway #3: Slowed growth is still growth

Source: Stanford HAI (AI Index, Chapter 4)

Stanford reported that “Global AI private investment was $91.9 billion in 2022, which represented a 26.7% decrease since 2021. The total number of AI-related funding events as well as the number of newly funded AI companies likewise decreased.”

This seems counterintuitive, especially with all the buzz around AI. Just last week there was news of OpenAI getting an $80 Billion valuation.

Source: State of AI 2023, Slide 115

The State of AI report put it differently. “GenAI saves the VC world. AI startups focused on generative AI applications, raised over $18 billion from VC and corporate investors.”

So, what’s the cohesive narrative? Well, the punchline is basically that the growth of AI investments is slowing.

But growth is still growth. To put it more concretely: The valuation of an e-commerce store netting $50k a day is no match for a pre-product AI tool that will “revolutionize” spreadsheets. 

🚀 Key takeaway #3: The “vibe check” on technological innovation

If you want a complete run-through of AI Benchmarks in 2023, check out this overview

Nevertheless, here’s what the Stanford Institute for Human-Centered Artificial Intelligence (HAI) has to say about AI benchmarking:

“AI continued to post state-of-the-art results, but year-over-year improvement on many benchmarks continues to be marginal. Moreover, the speed at which benchmark saturation is being reached is increasing. However, new, more comprehensive benchmarking suites such as BIG-bench and HELM are being released.”

Stanford HAI, “2023 AI Index Report”

Source: Stanford HAI (AI Index, Chapter 2)

In other words, AI is getting better, but the rate at which we’re improving may not match the rate at which the AI community is hyping things up on Twitter.

For more information on the history of benchmarks and where exactly our current benchmarks come from, check out this article

Oh, and one more thing: The State of AI Report conducted a “Vibe Check” on LLM leaderboards and benchmarks. Check out that analysis here (Slide 32).

Source: State of AI, Slide 32

🔊 Key Takeaway #4: Voice Technology is on the rise

Finally, let’s talk about voice technology.

It’s one of the most rapidly growing sectors of AI—both in terms of technological innovation and financial attention. And there are a few key points we’d like to make with respect to audio and voice.

  • Voice technology adoption is on the rise. It’s clear that as companies scale, the need for some type of voice technology grows. This year 82% of people in this space confirmed their current utilization of voice technology, a 6% increase from last year. In addition, 66% percent consider voice-enabled experiences to be a critical component of their organization’s future strategy, highlighting the growing dependence on voice technologies.

  • High Return on Investment. It’s no surprise that cost savings yielded the highest return on investment, with 51% of the survey participants indicating they saved between 26-75% by implementing voice technology. However, productivity gains aren’t far behind. 49% of respondents said that productivity increased between 26-75% percent as a direct result of their investment in voice technology.

  • The future of Speech AI. Considering the advances we’ve seen in conversational AI, the voice-enabled future may come even sooner. 43% of respondents believe voicebots will achieve human-like levels of interaction in less than a year, and 54% said that human-like AI voicebots are between one and three years away.

    Check out the full report here.

🌟 Get up to $100,000 in speech-to-text credits.
Join the Deepgram Startup Program 🚀