Where are we with AI now evolution

How is AI evolving

Introduction

I have written about AI from business perspective. Here I will look forward and cover how my original vision of AI as a business tool is coming along.

Artificial Intelligence was introduced to most of us as ‘smart chat’. My original belief however was shaped earlier in 2023 which took it much further than chat. The shift in my vision took the impact a holistic vision of business benefit, social change and structutal change. The model I had in mind is the British Industrial revolution which I covered earlier. There the shifts from structural change came first and eventually followed by GDP improvement at a scale otherwise impossible.

Analysis

Chat was a simple function which produced efficiency of that function, reducing manual intervention and anticipated most frequently asked questions. This was basic process automation and epitomised the expection of computerisation then. The concept was straight through processing of current processes. Chat was chosen because call centres were labour, technology and physical resource intensive

AI introduced lateral thinking developed by Machine learning. This when average people began to see the potential in AI yet most of us were still stuck on AI as process automation.

A breakthrough ocurred in the Transformer method developed by five Deepmind (ex google employees):

The best performing models also connect the encoder and decoder through an attention mechanism. We propose a new simple network architecture, the Transformer, based solely on attention mechanisms, dispensing with recurrence and convolutions entirely

This brought ingelligent thinking which now covers flields of medicine, engineering, finanncial services, philosophy, you name it. The transformer architecture is built on two key components: self-attention and positional encoding. Self-attention allows the model to focus on different parts of the input sequence, determining how much focus to put on each part when processing a particular word or element. This mechanism enables the model to understand context and relationships within the data

In a few short months the landscape has shifted.

First some examples, then deeper analysis of current state and where it is ultimately taking us. Productivity!

Examples of AI evolution

Component types

• For a complete LLM list of LLM refer to Hugging Face.
• Here are the main corporate players
• Meta — it appears self serving but represents a large and a significant contributor to institutional research
• Microsoft — While considered as leaders in investment in security and industry leadership they nevertheless have a disturbingly high frequency of security incidents.
• Google — Similarly large research and advanced search
• Perplexity — experimental — various LLM models, always changing. Significant investment from corporate players.

New players

• Anthropic
• Llama
• Gemini
• Codellama
• Mistral

LLM examples that specialise in particular areas.

Areas of Specialists

• Medicine and clinical is making enormous strides
• News: Bloomberg has made huge improvements with BloombergGPT. Their dataset has 50% composed of targeted news and Bloomberg terminology.
• Apple shifting to device based AI with less focussed on server
• LMNT — a zero-sugar electrolyte drink mix was created by a former research biochemist to deliver the optimal ratio of sodium, potassium and magnesium.
• Legal services
• Financial Services
• Marketing

Conclusion

here are some key points about the current state of AI from a business and social change perspective in 2024:

Business Impact

• AI adoption in businesses is rapidly increasing, with 73% of U.S. companies having adopted AI in at least some areas of their business as of 2023 . Generative AI like ChatGPT is leading this wave, with 54% having implemented it .
• Businesses are seeing tangible benefits from AI adoption like reduced costs (42% reporting this) and increased revenues (59%) . AI is boosting productivity and efficiency across functions like marketing, sales, product development and IT .
• However, lack of technical talent/skills (39%) and concerns around trust, risk and responsible AI practices remain barriers to greater AI adoption .
• Investment in generative AI skyrocketed in 2023, nearly octupling from 2022 levels to $25.2 billion as companies raced to develop these capabilities .
• AI is expected to increasingly transform labor markets, business models and industries in 2024 as adoption grows . Roles and skills requirements are shifting.

Social Impact

• AI integration into critical decision-making areas like healthcare, education and finance is growing but raising concerns around privacy, fairness, transparency and accountability .
• There is a significant lack of standardization currently in how AI developers test and report on responsible AI practices like privacy and bias mitigation .
• Public perception and trust in AI is shifting, with growing debates around ethical AI governance as real-world impacts increase .
• Governments are increasingly regulating AI, with the U.S. seeing a 56% rise in AI-related regulations from 2022–2023 aimed at issues like foreign trade and data rights .
• AI is being leveraged for both positive applications like assistive technologies, as well as negative ones like spreading disinformation and deepfakes, amplifying societal challenges.
• Organised crime including gangs and state actors are displaying rapid growth in AI usage. This is particularly dangerous

In summary, AI adoption shows businesses seeking efficiency, but responsible development practices are still lacking standardization as societal risks become more apparent, driving greater public scrutiny and government regulation in 2024