AI in 2024

Over the past few weeks, I had time to relax, recuperate and reflect on the breakout year AI has had in 2023 and where things are going to go in 2024.

I’m going to share my personal take about AI in 2024.

Undoubtedly, we are in the age of AI, the official launch of ChatGPT in November 2022, started 2023 on a wave of unprecedented hype, where ChatGPT could arguably be the fasted spreading new technology, reaching 100 million users in just 2 months.

This led to a proliferation of commercial applications making some very lofty claims. Technology waves work like this, there is an excitement factor associated with the new and promising, a rush of early adopters looking to gain a commercial advantage and then disillusionment creeps in as promises are unfulfilled.

What makes this technology wave truly revolutionary is that for the first time in history we are building new intelligence which is both exhilarating and terrifying at the same time.

As the hype settles, 2024 will be a year for recalibrating expectations and building trust.

Reliable, Trustworthy AI

The fear of missing out is seeing AI tools being deployed at pace with promises of incredible productivity boost. However, users are still reporting hallucinations, and unreliable or misleading content. The question of “can I trust this output?” will only become more prevalent over the course of 2024.

AI is not infallible, and businesses will have to establish a framework for understanding and managing new risks while maximising the utility of AI solutions. Understanding that AI systems are tools that require human oversight and supervision will make the adoption of such solutions easier in the long run.

I expect to hear the phrase “AI Literacy “more often as the skills and competencies required to use AI technology and applications effectively, ethically and within the bounds of regulatory and legal compliance, become essential everyday business requirements.

The Value of Data

The recent publicity and legal cases around intellectual property infringement and data exploitation are refocusing the conversation on the true value of data, i.e. who owns it and how it is being utilised.

AI giants like OpenAI are suspected of violating data protection laws in multiple jurisdictions with respect to how they have obtained and utilised training data for their foundational models. Essentially these models were trained on vast quantities of data scraped from the internet that included personal data on individuals.

For example, the Italian data protection authority temporarily blocked access to OpenAI’s ChatGPT model last year, suspecting it to be breaching Articles 5, 6, 8, 13 and 25 of the General Data Protection Regulation (GDPR). While OpenAI was able to resume service relatively quickly, the laundry list of violations has been under intense scrutiny and will return to the top courts this year.

There are big thorny questions around the legal processing of data and the potential for these tools to cause unjustified harm to individuals (spread disinformation and even defamatory opinions off illegally processed data).

It is not inconceivable that services could be suspended again on a larger scale while data ownership and usage is debated in courts.

AI Everywhere

While the legal processes continue at their own inexplicable pace, the cat is truly out of the bag as access to cutting edge models (via APIs) and the ability to fine-tune these models with your own data has never been easier.

Cloud-based scalable infrastructure and user-friendly tools (and tutorials) mean you don’t have to be a highly technical data scientist to build and deploy a custom Large Language Model (LLM) application.

Developers can now deploy their own applications in a matter of minutes, then iterate and scale their creations as required. Our own experiments are yielding surprising successes early and often in the development cycle. For example, our:

• “Chat with your data” application allows you to query your data via a natural language interface, ask detailed questions of your data, calculate metrics, generate analysis in real-time. It’s like having an ultra-fast, experienced data scientist in the palm of your hand.

• AI-powered CRM systems convert audio files into accurate transcripts, topic-based summaries and meeting minutes, making the tracking of decisions, key information, and action items a seamlessly automated process.

• AI for data quality, we’re not talking about detecting outliers though we have had major success with that. We are developing ‘self-healing data pipelines’, a fancy way of saying a process that detects and resolves data quality issues in-system. AI systems are perfect for curating and improving data quality, continuously learning, and improving data quality over time.

The challenges of moving from a prototype to a robust application aren’t trivial, but the pace of innovation and the availability of highly proficient tools means we will be seeing many new custom applications on the market throughout 2024.

What’s in store for 2024?

There is no crystal ball and it is impossible to say exactly how 2024 will play out but in my humble opinion:

• 2024 will be a reset year for our AI expectations, the hype will lessen and the real work of understanding and managing the impact of AI will begin.
• There will be a rush and an earnest need to educate ourselves, improve our AI literacy so we can understand this new technology and not get left behind.
• The legislative arena will be a combative one and pushed to the forefront. It doesn’t help that it is an election year in more than 60 countries worldwide (including the USA, UK & India) and the risks of persuasive disinformation or misinformation are the highest they have ever been.
• The march of progress will continue, relentlessly, and the AI technology stack will continue to mature and become integrated into existing systems becoming part of our daily lives.

Equipped AI is helping our clients, leading alternative investment managers, navigate the new Era of AI. If any of the topics above are of interest, please contact us – sales@equipped.ai