Age of artificial intelligence

The foundations for the Age of Artificial Intelligence were laid during the latter part of the 20th century and the early 2000s. Key developments included advancements in computer science, neural network models, data storage, the Internet, and optical networking, enabling rapid data transmission essential for AI progress.{{cite journal|last=Rashid|first=Adib Bin|last2=Kausik|first2=MD Ashfakul Karim|title=AI revolutionizing industries worldwide: A comprehensive overview of its diverse applications|journal=Hybrid Advances|volume=7|date=2024|doi=10.1016/j.hybadv.2024.100277|doi-access=free|page=100277}}

The transition to this new era is characterized by the ability of machines to process and store information, and also learn, adapt, and make decisions based on complex data analysis. This shift is significantly affecting various sectors, including healthcare, finance, education, transportation, and entertainment.{{cite book|last=Russell|first=Stuart|last2=Norvig|first2=Peter|title=Artificial Intelligence: A Modern Approach|edition=4th|publisher=Pearson Higher Education|publication-place=Hoboken, NJ|date=2019|isbn=978-0-13-461099-3|url=https://aima.cs.berkeley.edu/}}

Tegmark's book, Life 3.0: Being Human in the Age of Artificial Intelligence, details a phase in which AI can independently design its hardware and software, transforming human existence. He highlights views from digital utopians, techno-skeptics, and advocates for ensuring AI benefits humanity.{{cite web|last=Tegmark|first=Max|title=LIFE 3.0|website=Kirkus Reviews|date=2017-08-29|url=https://www.kirkusreviews.com/book-reviews/max-tegmark/life-30/|access-date=2025-02-11}}

Leopold Aschenbrenner, a former employee of OpenAI's Superalignment team, focused on improving human decision-making with AI. In June 2024, he outlined a phased progression from data processing to augmented decision-making, autonomous actions, and, ultimately, AI with holistic situational awareness.Aschenbrenner, Leopold (2023). Situational Awareness: Artificial Intelligence and Human Decision-Making.{{cite web|last=Varanasi|first=Lakshmi|title=A researcher fired by OpenAI published a 165-page essay on what to expect from AI in the next decade. We asked GPT-4 to summarize it.|website=Yahoo Tech|date=2024-06-10|url=https://www.yahoo.com/tech/researcher-fired-openai-published-165-184227878.html|access-date=2025-02-11}}

Sam Altman, founder of OpenAI, has predicted that AI will reach a point of superintelligence within the year 2025. Superintelligence was popularized by philosopher Nick Bostrom, who defines it as "any intellect that greatly exceeds the cognitive performance of humans" in his 2014 book Superintelligence: Paths, Dangers, Strategies.

Altman outlined a phased approach to AI development that began with AI's early, narrow focus on specific tasks, which then transitioned to general intelligence that aligns with human values and safety considerations.{{cite web|last=Pillay|first=Tharin|title=How OpenAI’s Sam Altman Is Thinking About AGI and Superintelligence in 2025|website=TIME|date=2025-01-08|url=https://time.com/7205596/sam-altman-superintelligence-agi/|access-date=2025-02-11}} The next phase is a collaboration between humanity and AI, and the final phase is superintelligence, in which AI must be controlled to ensure it is benefiting humanity as a whole.Altman, Sam (2024). OpenAI's Vision for Responsible AI Development. Altman also outlines five levels of AI capability growth from generative AI, cognition, agentics, and scientific discovery to automated innovation.Altman, Sam (2024). The Five Phases of AI Capabilities.{{cite web|last=Cook|first=Jodie|title=OpenAI’s 5 Levels Of ‘Super AI’ (AGI To Outperform Human Capability)|website=Forbes|date=2024-07-16|url=https://www.forbes.com/sites/jodiecook/2024/07/16/openais-5-levels-of-super-ai-agi-to-outperform-human-capability/|access-date=2025-02-11}}

American computer scientist and writer Ray Kurzweil predicts a path leading to what he refers to as "The Singularity" around 2045.{{cite book|last=Kurzweil|first=Ray|title=The Singularity Is Near|year=2005|publisher=Viking Books|location=New York|isbn=978-0-670-03384-3|url-access=registration|url=https://archive.org/details/singularityisnea00kurz}} ([https://web.archive.org/web/20140316213301/http://hfg-resources.googlecode.com/files/SingularityIsNear.pdf PDF]) His phases include substantial growth in computing power, narrow AI, general AI (expected by 2029), and lastly, the integration of human and machine intelligence.{{cite web|last=Corbyn|first=Zoë|title=AI scientist Ray Kurzweil: ‘We are going to expand intelligence a millionfold by 2045’|website=the Guardian|date=2024-06-29|url=https://www.theguardian.com/technology/article/2024/jun/29/ray-kurzweil-google-ai-the-singularity-is-nearer|access-date=2025-02-11}}Kurzweil, Ray (2024). Updated Predictions on AGI and the Singularity.

From 2019 to 2023, there was a significant jump in AI capabilities, exemplified by the progression from GPT-2 to GPT-4, which saw AI models advance from grade-school level to advanced high-school level capabilities. This progress is measured in orders of magnitude increases in computing power and algorithmic efficiencies.

In 2017, researchers at Google introduced the Transformer architecture in a paper titled "Attention Is All You Need," authored by computer scientist Ashish Vaswani, and others.{{cite web|last=Vaswani|first=Ashish|last2=Shazeer|first2=Noam|last3=Parmar|first3=Niki|last4=Uszkoreit|first4=Jakob|last5=Jones|first5=Llion|last6=Gomez|first6=Aidan N.|last7=Kaiser|first7=Lukasz|last8=Polosukhin|first8=Illia|title=Attention Is All You Need|website=arXiv.org|date=2017-06-12|url=https://arxiv.org/abs/1706.03762|access-date=2025-04-01}} Transformers revolutionized natural language processing (NLP) and subsequently influenced various other AI domains.{{cite web|last=Devlin|first=Jacob|last2=Chang|first2=Ming-Wei|last3=Lee|first3=Kenton|last4=Toutanova|first4=Kristina|title=BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding|website=arXiv.org|date=2018-10-11|url=https://arxiv.org/abs/1810.04805|access-date=2025-04-01}} Key features of Transformers include their attention mechanism, which allows the model to weigh the importance of different parts of the input data dynamically; their ability to process input data in parallel, significantly speeding up training and inference compared to recurrent neural networks; and their high scalability, allowing for the creation of increasingly large and powerful models.{{cite web|last=Brown|first=Tom B.|last2=Mann|first2=Benjamin|last3=Ryder|first3=Nick|last4=Subbiah|first4=Melanie|last5=Kaplan|first5=Jared|last6=Dhariwal|first6=Prafulla|last7=Neelakantan|first7=Arvind|last8=Shyam|first8=Pranav|last9=Sastry|first9=Girish|last10=Askell|first10=Amanda|last11=Agarwal|first11=Sandhini|last12=Herbert-Voss|first12=Ariel|last13=Krueger|first13=Gretchen|last14=Henighan|first14=Tom|last15=Child|first15=Rewon|last16=Ramesh|first16=Aditya|last17=Ziegler|first17=Daniel M.|last18=Wu|first18=Jeffrey|last19=Winter|first19=Clemens|last20=Hesse|first20=Christopher|last21=Chen|first21=Mark|last22=Sigler|first22=Eric|last23=Litwin|first23=Mateusz|last24=Gray|first24=Scott|last25=Chess|first25=Benjamin|last26=Clark|first26=Jack|last27=Berner|first27=Christopher|last28=McCandlish|first28=Sam|last29=Radford|first29=Alec|last30=Sutskever|first30=Ilya|last31=Amodei|first31=Dario|title=Language Models are Few-Shot Learners|website=arXiv.org|date=2020-05-28|url=https://arxiv.org/abs/2005.14165|access-date=2025-04-01}}

Transformers have been used to form the basis of models like BERT and GPT series, which have achieved state-of-the-art performance across a wide range of NLP tasks. Transformers have also been adopted in other domains, including computer vision, audio processing, and even protein structure prediction.{{cite journal|last=Jumper|first=John|last2=Evans|first2=Richard|last3=Pritzel|first3=Alexander|last4=Green|first4=Tim|last5=Figurnov|first5=Michael|last6=Ronneberger|first6=Olaf|last7=Tunyasuvunakool|first7=Kathryn|last8=Bates|first8=Russ|last9=Žídek|first9=Augustin|last10=Potapenko|first10=Anna|last11=Bridgland|first11=Alex|last12=Meyer|first12=Clemens|last13=Kohl|first13=Simon A. A.|last14=Ballard|first14=Andrew J.|last15=Cowie|first15=Andrew|last16=Romera-Paredes|first16=Bernardino|last17=Nikolov|first17=Stanislav|last18=Jain|first18=Rishub|last19=Adler|first19=Jonas|last20=Back|first20=Trevor|last21=Petersen|first21=Stig|last22=Reiman|first22=David|last23=Clancy|first23=Ellen|last24=Zielinski|first24=Michal|last25=Steinegger|first25=Martin|last26=Pacholska|first26=Michalina|last27=Berghammer|first27=Tamas|last28=Bodenstein|first28=Sebastian|last29=Silver|first29=David|last30=Vinyals|first30=Oriol|last31=Senior|first31=Andrew W.|last32=Kavukcuoglu|first32=Koray|last33=Kohli|first33=Pushmeet|last34=Hassabis|first34=Demis|title=Highly accurate protein structure prediction with AlphaFold|journal=Nature|volume=596|issue=7873|date=2021-08-26|issn=0028-0836|pmid=34265844|pmc=8371605|doi=10.1038/s41586-021-03819-2|doi-access=free|pages=583–589|url=https://www.nature.com/articles/s41586-021-03819-2.pdf|access-date=2025-04-01}}

Transformers face limitations, including quadratic time and memory complexity with respect to sequence length, lack of built-in inductive biases for certain tasks, and the need for vast amounts of training data.{{cite web|last=Tay|first=Yi|last2=Dehghani|first2=Mostafa|last3=Bahri|first3=Dara|last4=Metzler|first4=Donald|title=Efficient Transformers: A Survey|website=arXiv.org|date=2020-09-14|url=https://arxiv.org/abs/2009.06732|access-date=2025-04-01}}{{cite web|last=Battaglia|first=Peter W.|last2=Hamrick|first2=Jessica B.|last3=Bapst|first3=Victor|last4=Sanchez-Gonzalez|first4=Alvaro|last5=Zambaldi|first5=Vinicius|last6=Malinowski|first6=Mateusz|last7=Tacchetti|first7=Andrea|last8=Raposo|first8=David|last9=Santoro|first9=Adam|last10=Faulkner|first10=Ryan|last11=Gulcehre|first11=Caglar|last12=Song|first12=Francis|last13=Ballard|first13=Andrew|last14=Gilmer|first14=Justin|last15=Dahl|first15=George|last16=Vaswani|first16=Ashish|last17=Allen|first17=Kelsey|last18=Nash|first18=Charles|last19=Langston|first19=Victoria|last20=Dyer|first20=Chris|last21=Heess|first21=Nicolas|last22=Wierstra|first22=Daan|last23=Kohli|first23=Pushmeet|last24=Botvinick|first24=Matt|last25=Vinyals|first25=Oriol|last26=Li|first26=Yujia|last27=Pascanu|first27=Razvan|title=Relational inductive biases, deep learning, and graph networks|website=arXiv.org|date=2018-06-04|url=https://arxiv.org/abs/1806.01261|access-date=2025-04-01}}{{cite web|last=Kaplan|first=Jared|last2=McCandlish|first2=Sam|last3=Henighan|first3=Tom|last4=Brown|first4=Tom B.|last5=Chess|first5=Benjamin|last6=Child|first6=Rewon|last7=Gray|first7=Scott|last8=Radford|first8=Alec|last9=Wu|first9=Jeffrey|last10=Amodei|first10=Dario|title=Scaling Laws for Neural Language Models|website=arXiv.org|date=2020-01-23|url=https://arxiv.org/abs/2001.08361|access-date=2025-04-01}}{{cite web|last=Fournier|first=Quentin|last2=Caron|first2=Gaétan Marceau|last3=Aloise|first3=Daniel|title=A Practical Survey on Faster and Lighter Transformers|url=https://dl.acm.org/doi/10.1145/3586074|website=acm.org|date=2021-03-26|doi=10.1145/3586074|doi-access=free}} The complexity of Transformer models also often makes it challenging to interpret their decision-making processes.{{cite web|last=Vig|first=Jesse|title=A Multiscale Visualization of Attention in the Transformer Model|website=arXiv.org|date=2019-06-12|url=https://arxiv.org/abs/1906.05714|access-date=2025-04-01}}

To address these limitations, researchers are exploring various approaches, including alternative attention mechanisms (Reformer, Longformer, BigBird), sparse attention patterns, Mixture of Experts (MoE) approaches, and retrieval-augmented models.{{cite web|last=Lewis|first=Patrick|last2=Perez|first2=Ethan|last3=Piktus|first3=Aleksandra|last4=Petroni|first4=Fabio|last5=Karpukhin|first5=Vladimir|last6=Goyal|first6=Naman|last7=Küttler|first7=Heinrich|last8=Lewis|first8=Mike|last9=Yih|first9=Wen-tau|last10=Rocktäschel|first10=Tim|last11=Riedel|first11=Sebastian|last12=Kiela|first12=Douwe|title=Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks|website=arXiv.org|date=2020-05-22|url=https://arxiv.org/abs/2005.11401|access-date=2025-04-01}} Researchers are also exploring neuro-symbolic AI and multimodal models to create more versatile and capable AI systems.{{cite web|last=Lamb|first=Luis C.|title=Neurosymbolic AI: The 3rd Wave|website=arXiv.org|date=2020-12-10|url=https://arxiv.org/abs/2012.05876|access-date=2025-04-01}}{{cite web|last=Ramesh|first=Aditya|last2=Dhariwal|first2=Prafulla|last3=Nichol|first3=Alex|last4=Chu|first4=Casey|last5=Chen|first5=Mark|title=Hierarchical Text-Conditional Image Generation with CLIP Latents|website=arXiv.org|date=2022-04-13|url=https://arxiv.org/abs/2204.06125|access-date=2025-04-01}}

Optical networking is fundamental to AI system functioning. Optical fiber and laser technologies, such as dense wave division multiplexing power all the optical networks that enable the collection, updating, processing, and transmission of vast datasets used for training AI models. Data centers store the processed data required by users of Large Language Models (LLMs) and other AI applications.{{cite journal|last=Wu|first=Jiamin|last2=Lin|first2=Xing|last3=Guo|first3=Yuchen|last4=Liu|first4=Junwei|last5=Fang|first5=Lu|last6=Jiao|first6=Shuming|last7=Dai|first7=Qionghai|title=Analog Optical Computing for Artificial Intelligence|journal=Engineering|volume=10|date=2022|doi=10.1016/j.eng.2021.06.021|doi-access=free|pages=133–145}}{{cite journal|last=Nevin|first=Josh W.|last2=Nallaperuma|first2=Sam|last3=Shevchenko|first3=Nikita A.|last4=Li|first4=Xiang|last5=Faruk|first5=Md. Saifuddin|last6=Savory|first6=Seb J.|title=Machine learning for optical fiber communication systems: An introduction and overview|journal=APL Photonics|volume=6|issue=12|date=2021-12-01|issn=2378-0967|doi=10.1063/5.0070838|doi-access=free|url=https://aip.scitation.org/doi/pdf/10.1063/5.0070838|access-date=2025-04-01|page=}}

By 2030, data transmission volumes are expected to increase by more than ten times compared to 2020 levels. This growth is accompanied by advancements in data processing technologies, including the development of quantum-sensing technologies and massive data centers.{{cite web|last=Tavakoli|first=Asin|last2=Harreis|first2=Holger|last3=Rowshankish|first3=Kayvaun|last4=Bogobowicz|first4=Michael|title=Charting a path to the data- and AI-driven enterprise of 2030|website=McKinsey & Company|date=2024-09-05|url=https://www.mckinsey.com/capabilities/mckinsey-digital/our-insights/charting-a-path-to-the-data-and-ai-driven-enterprise-of-2030|access-date=2025-04-01}}

Generative AI has emerged as a transformative technology in the Age of Artificial Intelligence.{{cite web|last=Kocher|first=Chris|title=Work in the age of artificial intelligence|website=News - Binghamton University|date=2025-04-01|url=https://www.binghamton.edu/news/story/5194/work-in-the-age-of-artificial-intelligence|access-date=2025-04-01}} As of 2025, 75% of organizations surveyed by McKinsey & Company reported they regularly use generative AI, which is an increase of 10% from the previous year. Companies have largely worked with the 10% of data that is structured, including transactions, SKUs, and product specifications. However, generative AI is now providing access to the remaining 90% of unstructured data, which includes videos, images, chats, emails, and product reviews.

The Age of Intelligence is expected to have profound economic implications as AI could contribute up to $19.9 trillion to the global economy by 2030.{{cite web|last=Fried|first=Ina|title=AI is predicted to add $19.9 trillion to the global economy through 2030, per IDC|website=Axios|date=2024-09-17|url=https://www.axios.com/2024/09/17/ai-global-economy-idc-2030|access-date=2025-04-01}} This economic transformation is anticipated from increased productivity, automation of cognitive tasks, and the creation of new industries and job categories.{{cite web|title=The Future of Jobs Report 2020|website=World Economic Forum|date=2020-10-20|url=https://www.weforum.org/publications/the-future-of-jobs-report-2020/|access-date=2025-04-01}}

The rise of AI and automation technologies is leading to significant changes in the workforce. While there are concerns about job displacement, many specialists argue that AI will create new job categories and drive productivity growth. New roles such as prompt engineers, AI ethics stewards, and unstructured-data specialists are emerging.

AI-powered drug discovery could generate up to $70 billion in savings for the pharmaceutical industry by 2028.{{cite web|title=Artificial Intelligence in Drug Discovery Market Size Analysis and Opportunity, 2018-2028|website=Bekryl Market Analysts|date=2018-06-11|url=https://bekryl.com/industry-trends/ai-artificial-intelligence-in-drug-discovery-market-size-analysis|access-date=2025-04-01}}

The global autonomous vehicle market is projected to reach $556.67 billion by 2026.{{cite web|last=Garsten|first=Ed|title=Sharp Growth In Autonomous Car Market Value Predicted But May Be Stalled By Rise In Consumer Fear|website=Forbes|date=2018-08-13|url=https://www.forbes.com/sites/edgarsten/2018/08/13/sharp-growth-in-autonomous-car-market-value-predicted-but-may-be-stalled-by-rise-in-consumer-fear/|access-date=2025-04-01}} Leveraging the mobile telephone infrastructure, AI-powered traffic management systems can reduce urban travel times by up to 20%.{{cite web|last=Woetzel|first=Lola|last2=Remes|first2=Jaana|last3=Boland|first3=Brodie|last4=Lv|first4=Katrina|last5=Sinha|first5=Suveer|last6=Strube|first6=Gernot|last7=Means|first7=John|last8=Law|first8=Jonathan|last9=Cadena|first9=Andres|last10=Tann|first10=Valerie von der|title=Smart cities: Digital solutions for a more livable future|website=McKinsey & Company|date=2018-06-05|url=https://www.mckinsey.com/capabilities/operations/our-insights/smart-cities-digital-solutions-for-a-more-livable-future|access-date=2025-04-01}}

{{Portal|Technology}}

• Machine learning
• Deep learning
• Big data
• Internet of Things (IoT)
• Quantum computing
• Ethics of artificial intelligence
• Digital transformation

{{Reflist}}

Category:History of artificial intelligence

Category:Technology in society

Category:2010s in computing

Category:2020s in computing

Category:Max Tegmark