Predictions, Predictions Everywhere

Data Unfiltered

Predictions for the New Year

It’s the start of the new year, and with that come predictions from everywhere.

Here are my 10 predictions for 2024:

1. Lawsuits, lawsuits, and more lawsuits.

2. Investments cut for building in-house AI solutions.

3. Increased focus on enterprise gen-AI tools.

4. Decrease in the amount of gen-AI companies.

5. Continued shift toward personalized gen-AI.

6. SLMs take center stage.

7. CAIO positions move from rarity to semi-common.

8. Increase in the number of CDOs.

9. More fractional data leadership in the world.

10. Companies shift back towards On-Prem.

Lawsuits, lawsuits, and more lawsuits.

Starting with the most exhilarating topic—lawsuits.

I anticipate the New York Times lawsuit against Microsoft and OpenAI is only the beginning of what will come.

While I don’t expect much from this initial lawsuit, it will serve as a precedent for many future lawsuits. There will likely be class action suits from artists, writers, journalists, and others who have had their data used to build these models by OpenAI, Google, Microsoft, Meta, X, and others.

Entire industries also feel threatened by the prospect of gen-AI and will see litigation as a viable method to hinder or stop AI from evolving further.

The SAG-AFTRA and Writers’ Strike are prime examples of industries and individuals feeling threatened by the rise of generative AI and artificial intelligence. While agreements were met with both unions, more companies, like Meta, will allow creators and celebrities to license their likenesses, further blurring the line between acceptable use of derivative works and monetizing likenesses.

Investments cut for building in-house AI solutions.

The hype train for 2023 was unequivocally generative AI.

Every company bought into the idea that they need to build in-house generative AI capabilities. Otherwise, they will be far behind all competitors and lose revenue, market share, and eventually their company without immediate innovation.

Unfortunately, most companies are finding that they lack the proper data foundation to build generative AI, which was the same story for AI 2+ years ago.

Suppose you don’t have the correct data or an understanding of your data. In that case, it doesn’t matter how fantastic or innovative your proof-of-concepts are—they won’t go anywhere without proprietary data.

This is why the “build” de’ “build” decision will be at a loss, and they will turn to enterprise solutions that better fit their needs and timelines.

Increased focus on enterprise gen-AI tools.

For the most part, many of the gen-AI use cases that have been developed have focused on individual and retail use. Yes, there are tools, like GitHub Copilot, that have been developed, but the primary focus has been on the individual.

The primary reason is that many companies have been wary of adopting gen-AI tools, given the potential security concerns, risks of leaking confidential and proprietary information, and recent cyberattacks.

As these tools mature and exit beta and testing phases, many will likely obtain the appropriate SOC2, HITRUST, or other compliance certifications to gain the trust of organizations to incorporate these tools across the enterprise.

Decrease in the amount of gen-AI companies.

Since 2022, over 250 generative AI companies have been launched.

Given the constricting funding environment, lack of data foundations or proprietary data, and potential lawsuits, smaller companies will run out of funds and be forced to sell or close down.

The 250+ will likely decrease to 20-30% of the original number.

Continued shift toward personalized gen-AI.

The moat that big tech has around generative AI compared to these smaller companies is the years and years worth of data they have collected on each of us.

Instead of repeatedly optimizing prompts to obtain your desired result, what if you could incorporate your entire Google search history, Instagram feed, or transaction history so that the tool isn’t sending you a general response but rather a response tailored to your specific wants and needs?

Personalized generative AI will be the gen-AI 3.0 that gives everyone their own subconscious AI assistant.

SLMs take center stage.

I’ve long criticized the number of parameters needed to train and power these AI models.

Obviously, more parameters = better results, but this is a costly and unsustainable approach for the future.

To maintain these models, you need:

1. Better chips and hardware (like Sam Altman is attempting to create and fund).

2. To operate at a loss (like OpenAI does) until eventually raising prices across the board.

Instead of these two prohibitive paths, small language models (SLM) could save the day.

Microsoft and Apple have recently released SLMs, Phi-2, and Ferret, which perform the same or better than LLMs at specific tasks with a fraction of the parameters. This means the language models will likely be able to be run locally and not experience the same hardware or cost constraints.

I view SLMs and enhanced architectures as the most prudent path forward for generative AI.

CAIO positions move from rarity to semi-common.

Already announced by cio.com as the “hottest new job” Chief AI Officers will go from rarity to semi-common within organizations as companies attempt to build and leverage AI solutions.

Let’s hope that the CAIO role, unlike the CDO role, will be clearly defined and the right individuals will be hired instead of defaulting to a random technologist at the company.

Increase in the number of CDOs.

Given the data issues companies continue to face and the increased reliance on data for organizations, there will also be an increase in Chief Data Officer (CDO) roles in 2024.

CDOs have been poorly defined for years, which has led to the wrong people being hired for these roles and an average tenure (2.5 years) that is half of the other c-suite tenures (5 years).

Given the desire to build more advanced solutions and the data required to get there, more organizations will find the need to bring on a CDO to develop their foundational data layer.

More fractional data leadership in the world.

Many organizations have fewer C-suite positions because of the cost of having many executives with higher salaries, stock options, equity, benefits, etc.

A phenomenal alternative to hiring a full-time executive is hiring a fractional leader who can focus on a narrower scope.

Concerning data, a fractional data leader can help an organization:

1. Uncover existing pain points that data can support.

2. Create a list of prioritized use cases, starting with low-effort/high-ROI initiatives.

3. Create data strategies and roadmaps for future state planning.

4. Interview, hire, and mentor full-time operators.

   + more depending on the organization’s needs.

The leader would provide the organization with full-time value at a fraction of the cost without requiring other benefits that an FTE would receive.

Companies shift back towards On-Prem.

Many companies were sold on the idea that migrating to the cloud would save enormous costs.

Unfortunately, that is not always the case.

A company migrates to the cloud, but then they:

1. May be locked into the vendor who can raise prices.

2. Don’t have the right staff to handle the complexities of the new environment.

3. Migrated existing tech debt, which incurs compute and orchestration costs.

4. Need a suite of tools to manage the cloud, each with its own 5- to 6-figure bill.

Before you know it, you have a worse solution with 2-3x the price tag.

Given this, as organizations begin to read the fine print and see their high costs, many will shift back to On-Prem or take a hybrid approach instead of being fully cloud-dependent.

Data Career Tips

Stop Waiting and Take Action

While planning and strategizing effectively is essential, some use the guise of “perfection” to disable themselves and never take action.

Often, it’s better to execute and then refine instead of always trying to put on the “finishing touches” and getting stuck in analysis paralysis.

If you’ve been waiting to:

• Start your side hustle.

• Apply for that new job.

• Advocate for a new role at your company.

Make 2024 the year you achieve those goals and continue pushing forward.

Datinuumber of the Week: 2024

Happy New Year to all!

I wish everyone a happy, safe, and prosperous new year filled with phenomenal data points.

Thank You

Thank you for subscribing and reading this week’s newsletter.

If you enjoyed the newsletter, the best way to help is by sharing it with colleagues and friends.

If you prefer to listen to the newsletter, the Datinuum Podletter will be released weekly on Apple, Spotify, or your favorite podcast player.

Feedback is a gift. Please reach out to [email protected] with any feedback or questions.