AI is dominating conversations across industries, governments, and people. Let’s look at the trends that will prevail in 2024.
Responsible AI guardrails: Dialogues regarding AI safety and regulations will gain momentum and superpowers will introduce comprehensive AI laws. Enterprises will start building/adopting technical guardrails and regulation frameworks to ensure the safe use of AI. Industry bodies will develop new commercial standards for AI development (akin to ISO) and a new industry will emerge for auditing/certifying AI. It will open a vibrant commercial market for AI safety technical solutions and guardrails. We can expect rapid technological advancements in deepfake detection and encryption.
AI security: Security vulnerabilities of AI models are becoming visible, with threat vectors such as poisoning, inference attacks, and evasion attacks emerging that can potentially damage mission-critical AI. Attackers can manipulate the behaviour of AI models by poisoning their training data or administering controlled inputs. They can nullify model outputs by exploiting its vulnerabilities. We will see more market solutions and newer techniques available to hackers.
Multi-modal generative AI: Multi-modal Gen AI is gaining traction, opening new possibilities in customer service, personalisation, content creation, and more. By combining multiple input and output modalities, newer applications are unlocked. Gen AI assistants will augment human capabilities phenomenally and integrate more seamlessly into our lives.
Host of specialised models from foundational models: AI adoption is evolving from personal to specialised and custom AI applications using closed models and APIs. As Gen AI gets democratised, enterprises will create narrow Gen AI by fine-tuning the foundational generative AI open models on specific enterprise data to create custom applications. The focus will shift to industry-specific AI applications using specialised pre-trained models to deliver exceptional accuracies in specific domains or tasks.
Quantum AI: This refers to the intersection of quantum computing and AI. It uses quantum mechanics principles to enhance AI algorithms and solve complex problems more efficiently than classical computers. Quantum computing leverages the unique properties of quantum bits, or qubits, to perform computations that would be challenging for classical computers. The use of Quantum AI will revolutionise AI computing, and it will be exciting to see how the quantum realm unlocks more AI capabilities.
Artificial General Intelligence: This represents a form of AI that can understand, learn, and apply knowledge across different tasks at a level comparable to human intelligence. While current AI systems are specialised and excel at specific tasks, AGI aims to exhibit a level of versatility and generalisation akin to human cognitive abilities. We can expect rapid democratisation of autonomous AI agents and platforms like Super AGI and Baby AGI. The boundaries of what is possible with AI will expand, and the pace of adoption will also increase.
Growth of AI cloud and Poly AI platforms: This is encouraging enterprises to shift from point use cases to enterprise-wide AI deployment. They are developing abstraction layers to select and integrate AI providers, models, micro-AI platforms, and tooling as per their requirements. Enterprises will have more choice in platforms and infrastructure to build their AI capabilities according to their goals, preferences, and existing infrastructure.
The writer is Executive Vice President – Global Head of AI and Industry Verticals, Infosys