OpenAI eyes the wearables business: Robots, headsets, watches and a whole lot more - Related to evolve,, late, quantum-safe, as, their

If you're not working on quantum-safe encryption now, it's already too late

Remember Nokia? Back before smartphones, many of us carried Nokia's nearly indestructible cell phones. They no longer make phones, but don't count Nokia out. Ever since the firm was founded in 1865, Nokia has successfully pivoted to industries showing promise.

Here's a fun trivia fact you can use at your next party: Nokia once made toilet paper. In fact, the firm was initially founded as a pulp mill. Later, the Finnish firm made rubber boots and respirators.

Also: How AI will transform cybersecurity in 2025 - and supercharge cybercrime.

Here's another name you might be familiar with: Bell Labs. For years, Bell Labs was at the forefront of technology research. In fact, UNIX (which inspired Linux) was developed at Bell Labs, along with many other critical technologies like lasers, transistors, the C and C++ programming languages, and even optical fiber systems. In 2016, Nokia acquired Bell Labs.

Martin Charbonneau, head of Quantum-Safe Networks at Nokia Nokia.

Now, Nokia's portfolio of hardware and software solutions -- spanning mobile and fixed network infrastructure, cloud data center technologies, and beyond -- serves as a foundation for digitalization and the AI and quantum era across industries.

, head of Quantum-Safe Networks at Nokia, "7 out of 10 fiber-connected homes in the US use Nokia technology, 15 out of 20 power utilities in the US, and more than 1,000 public sector organizations worldwide trust our technologies for their critical operations."

Also: Google's quantum breakthrough is 'truly remarkable' - but there's more to do.

ZDNET had the opportunity to sit down with Martin to discuss another transformative technology on the cusp: quantum computing. Quantum computing is expected to be able to solve some problems a million times faster (yes, you read that right, a million) than conventional computing. Some of our most robust encryption algorithms could take tens or hundreds of thousands of years to crack using traditional computing. But with quantum computing, those problems could be solved in seconds.

Let's dive deep into what this all means for telecommunications, security, AI, and our future.

ZDNET: How does quantum computing differ from classical computing?

Martin Charbonneau: Conventional computers are based on the concept that electrical signals can be in only one of two states or binary bits to store and process data -- on or off, zeros and ones.

Quantum computers are based on the principles of quantum mechanics. Quantum computers can encode more data concurrently using quantum bits, or qubits, in superposition, which can scale exponentially. A qubit can behave like a bit and store either a zero or a one, but it can also be a weighted combination of zero and one at the same time.

Because they are not limited to only one state at a time, they can perform tasks exponentially faster than classical computers and can also carry out multiple processes at once, further increasing their capacity and speed.

ZDNET: Why does quantum computing pose such a significant threat to current encryption methods?

MC: Quantum computers can solve problems or compromise mathematical cryptography algorithms in mere minutes that would have taken even the biggest conventional supercomputers thousands of years to compromise.

The point when a quantum computer exists that can break common encryption in use today is called Q-Day, and the computer that could break it is referred to as a CRQC or Cryptographically Relevant Quantum Computer.

ZDNET: Could you provide an example of a critical industry particularly vulnerable to quantum-based attacks?

MC: Many of the particularly vulnerable industries are the organizations we think of as being targets of cyber threats today, like governments and defense organizations.

But in reality, with today's public key cryptography rendered useless, all networks -- across all industries -- will become vulnerable to attack. Threat actors could cripple critical infrastructure by attacking the networks that support them.

Also: How AI will transform cybersecurity in 2025 - and supercharge cybercrime.

Quantum threats could impact power and water supplies, public transportation systems, telecommunications, public safety communications, financial market data and systems, healthcare research and hospital networks, and more -- with life-threatening and economy-impacting consequences.

Quantum attacks won't target only those companies or organizations that are using quantum computers themselves. A CRQC poses a threat to any industry, as well as the businesses and individuals they serve.

It is a matter of risk management for all.

ZDNET: What are the primary encryption methods at risk with the advent of quantum computing?

MC: As we move into the Quantum [website] age [actual use, rather than theoretical research -- DG], many of the standard cryptography algorithms and protocols in place today are at risk from a CRQC.

The Information Communications Technology (ICT) industry is realizing the seismic impact of this and is undergoing a significant migration of its cryptographic practices, with many organizations already in the planning stage, and some in a migration or execution phase.

To date, we have been 'lucky' that our existing mathematics cryptography algorithms have not been previously compromised. So, moving forward we must build a robust and resilient cryptography tool kit that addresses the potential of quantum computing.

This is essential to ensure we can support our continued digitalization and ensure a Quantum Secure Economy.

ZDNET: What role does artificial intelligence play in both enabling and mitigating risks related to quantum computing?

MC: AI can significantly enhance quantum computing by optimizing quantum algorithms and improving efficiency. This means quantum computers can solve complex problems faster and more effectively by using fewer quantum computer resources. AI also helps in developing new quantum algorithms and managing the vast amounts of data processed by quantum computers.

On the flip side, AI may also enable quantum threats. For example, AI may help quantum computers break current encryption methods much faster with new algorithms. Additionally, AI may automate and enhance attack strategies, creating new ways to exploit vulnerabilities.

Also: ChatGPT's Deep Research just identified 20 jobs it will replace. Is yours on the list?

AI may also play a crucial role in defending against quantum threats. It may help develop quantum-safe cryptographic algorithms that are resistant to quantum attacks. AI-driven risk assessment tools may continuously monitor systems for potential threats, detect anomalies, and provide real-time insights to mitigate risks. This may enrich the security and trust of our digital infrastructure.

ZDNET: How imminent is the threat of quantum computers breaking existing encryption standards?

MC: The arrival of a CRQC is not an "if," it's a "when." The timing of a CRQC is directly related to the advancement (and stability) of quantum computing. The faster a mature/stable quantum computer arrives, the sooner the threat arrives.

There are many organizations and governments around the world working on advancing quantum computing technologies so we can realize the vast benefits of the technologies. Concurrently, other organizations are looking at the innovation speed and advancements to measure how soon a threat could arise.

One research on the topic is the Quantum Threat Timeline research from the Global Risk Institute . Their latest analysis puts a 14% chance of a CRQC becoming available in the next 5 years.

This may sound like a small number, but it increases rapidly with time, where the risk is over 60% in 15 years based on the current status of quantum computing. The pace of innovation in quantum computing is not slowing either. Its acceleration could mean the timeline looks different next year. So, the idea is to be aware of the threat and take action now to protect critical infrastructure.

Also: Why some companies are backing away from the public cloud.

While the availability of the CRQC may not come in the near term, threat actors are already preparing for Q-Day. Many are collecting encrypted data from target organizations today and storing it so that it can be decrypted when the evolution of quantum computing delivers a CRQC capable of rendering some existing cryptographic algorithms obsolete. The industry refers to this ongoing activity as harvest now, decrypt later (HNDL).

These are severe risks, and the timeline to transition to a new quantum computer-secure future, with techniques such as post-quantum cryptography security models, is intricate. Our industry must take proactive measures now. We need to plan and deploy quantum-safe cryptography-based solutions in a defense-in-depth approach to provide secure and trusted connectivity, enable a quantum-safe global economy, and continue digital transformation.

Many global policy, regulatory, and government agencies (CISA, NSA, NIST in the US, for example) are urging critical infrastructure industries to make the move now to protect their data and critical communications.

ZDNET: What is post-quantum cryptography?

MC: Post-quantum cryptography (PQC) is one of the key methods to protect sensitive information as quantum computers evolve, posing risks to current encryption.

By developing quantum-resistant algorithms, PQC helps ensure long-term data security and maintain trust in digital economies. PQC will be used in applications such as banking transactions, secure communications, and protecting intellectual property, with organizations like NIST in the United States leading standardization efforts.

Today, many applications rely on public key infrastructure (PKI) for the generation and management of encryption keys. PQC seeks to improve upon today's cryptography by modifying the underlying mathematical methods used by these ciphers. PQC is only one of the required elements in creating quantum-safe networks.

ZDNET: What role does standardization play in preparing industries for a quantum-secure future?

MC: For main principles or technologies, quantum security encompasses more than just post-quantum cryptography (PQC). It involves building cryptographic resiliency through a defense-in-depth approach, which we believe is realized by utilizing multi-layer encryption and diverse cryptosystems, such as pre-shared keys and quantum key distribution.

Meanwhile, standardization plays a critical role in preparing industries for a quantum-secure future by ensuring interoperability, security, and compliance. In the US, NIST's post-quantum cryptography (PQC) standards provide robust encryption algorithms designed to withstand quantum attacks. The IETF is integrating PQC algorithms into secure protocols, which are then adopted by 3GPP for telecommunications.

Globally, ETSI and ITU focus on Quantum Key Distribution (QKD) to secure communication networks. Additionally, cybersecurity recommendations from agencies such as the NSA, ANSSI, and BSI guide industries in adopting secure-by-design principles and quantum-resistant technologies.

These efforts collectively build a resilient and secure digital infrastructure, ready to face the challenges posed by quantum computing.

ZDNET: How are different industries preparing for quantum risks?

MC: Government and defense industries are on top of the risk and acting as leaders. We also see progressing adoption across other industries, like Banking Financial Services and Insurance (BFSI) and mission-critical networks.

Different industries move at different paces based on their risk profile and the complexity and criticality of their infrastructure. We see in virtually every industry we work with (which spans telecoms, the public sector, and enterprise) that some organizations are still in a learning phase, some are identifying their unique risks, and yet some are still in the assessment phase.

Some leading organizations (across different industries, interestingly) are engaging in partnerships to drive quantum-security. For many industries, movement will inevitably come as global policy, regulatory, and government agencies impose mandates to ensure quantum security.

ZDNET: How does Nokia's approach to quantum safety address the specific needs of these industries?

MC: As we continue on our digitalization journey, it's clear that the importance of having safe and trusted connections will only continue to grow. Our reliance on safe and trusted connectivity is increasing, and it's essential that we act now to shield our digital future from the quantum paradigm shift.

In addition to promoting the adoption of PQC for obtaining quantum-safe applications, we are also promoting quantum-safe networks. This focuses on agile solutions with a defense-in-depth approach, through multi-layered network cryptography technology options, that can adapt to unique business needs, deliver the confidence to scale network deployments, and evolve as the quantum threat evolves. This complementary approach is all about reducing risk and ensuring trust in our digital communication infrastructure.

We believe this outcome is not just a short-term solution, but a long-term strategy that will persist through time. It's a trust-enabling bridge between current networks and the future quantum economy. And it's not just about today -- it's about generations to come.

Consumers, enterprises, mission-critical infrastructure builders, and communication service providers are all seeking this outcome of having quantum-safe security. They want to ensure that their digital communication infrastructure and data remain secure, reliable, and trustworthy.

At Nokia, we're committed to delivering this outcome. We have quantum-safe solutions today -- proven and ready for immediate implementation. Concurrently, Nokia Bell Labs is at the forefront of leading-edge research in specific technological domains, driving innovation with key academic and technology partners and shaping the future of quantum computing and quantum-safe network solutions.

ZDNET: How does proactive quantum-safe planning compare in cost and effort to reactive measures taken after vulnerabilities are exploited?

MC: We've seen the effects and costs of significant cyber breaches. IBM has estimated in a findings that the cost of the average cyber breach is over $[website] USD. And even beyond the cost, the loss of public trust, and impact on a organization's brand can be significant.

To assess an organization's risk factor, Dr. Michele Mosca of the University of Waterloo and EvolutionQ created a risk assessment theorem. This is where an organization needs to take into consideration the time it will take for a CRQC to become a reality, the time it will take the organization to migrate its cybersecurity systems, and the length of time its data needs to remain secure.

Our industry needs to reflect on the time required to migrate to quantum-safe cryptography through the lens of the Mosca Equation, which further reinforces that we already have a zero-day vulnerability.

Conducting a cryptography migration in a crisis is far from ideal. Haste could create new vulnerabilities or incremental vulnerabilities, costs will be increased, and so forth. There's an opportunity to plan for this now, conduct a thorough, thoughtful migration strategy, and roll it out in an effective, controlled, and properly managed way.

ZDNET: How far along is quantum-safe encryption?

MC: There is an awakening in the industry. While PQC is currently in the news, there are other forms of quantum-safe cryptography, like Pre-Shared Key technology (which is actively available and deployed). They're evolving.

These technologies are mature and can be utilized now in a multi-layered approach to protect critical systems. QKD technology is also emerging, evolving, and becoming available.

Also: For anonymous browsing, these extensions are the next best thing to Tor.

The announcement of NIST standardization of PQC algorithms was discussed in this recent article from Nokia and Nokia Bell Labs.

ZDNET: How does the concept of "crypto-agility" fit into long-term planning for quantum resilience?

MC: Crypto-agility is the ability to quickly adapt to new cryptographic algorithms and protocols as threats evolve. We believe that crypto-agility is one of the essential components [of quantum resilience], but not the only one.

For enterprise applications, this means migrating over time from traditional Public Key Cryptography (PKC) methods such as RSA, which are vulnerable to quantum attacks, to Post-Quantum Cryptographic (PQC) algorithms.

However, crypto-agility is not just about migrating to new algorithms; it's also about the ability to adapt to new threats and vulnerabilities as they emerge. This flexibility ensures that our systems can seamlessly transition to stronger security measures without significant disruptions, maintaining robust protection against emerging vulnerabilities.

Crypto agility needs to be complemented with crypto-resiliency, which involves relying on a digital fabric of complementary quantum-safe cryptosystems. By integrating multiple cryptographic methods, including symmetric cryptography, we ensure continuous protection and adaptability, even in the face of advanced quantum threats.

This resilience is crucial for maintaining the integrity and security of our data over time. Should a PQC algorithm weaken or break over time, the other symmetric cryptosystem would still be offering protection.

Multi-layered quantum-safe cryptography adds additional layers of security by employing multiple quantum-resistant cryptographic techniques. For service providers and enterprises building network-layer connectivity, this means activating complementary quantum-safe network-level encryption using symmetric-based cryptography.

This approach complements the application layer, which uses PKC PQC-based cryptography, reducing the risk of a single point of failure and ensuring that if the application layer is compromised, others remain intact to provide ongoing protection.

Together, these strategies form a robust defense-in-depth framework. By combining crypto-agility, crypto-resiliency, and multi-layered quantum-safe encryption, we create a comprehensive and proactive security posture that can withstand current and future threats, ensuring the security and resilience of our digital infrastructure.

ZDNET: Are there challenges in integrating quantum-safe encryption into legacy systems, and how can they be overcome?

MC: The WEF has estimated that the quantum-safe cryptography migration could force the replacement of between 10 and 20 billion devices globally. Many of these devices are IoT devices and are not capable of migration to quantum-safe cryptography.

In terms of networks where Nokia is a key supplier, we've already embedded quantum-safe encryption engines into our product platforms and silicon.

The challenge for the networking industry is around the generation and automated generation, distribution, and deployment of quantum-safe cryptographic keys.

MC: Quantum-safe data protection complements these regulations. Whether data is in-flight, at rest, or during processing, ensuring data privacy and protection against emerging quantum threats is key to compliance.

ZDNET: Where will quantum-safe cryptography be used?

MC: Quantum-safe cryptography, in the context of our answers, mainly applies to the protection of data in flight.

It will also be applied to digital signatures, firmware, software downloads, etc., used in numerous use cases, from cloud access and data center interconnects, to the digital supply chain and more.

Quantum-safe measures will be integrated and aligned with broader cybersecurity, so at some point, we believe the aim is that everything will be quantum-safe.

ZDNET: What collaborative efforts between private companies and research institutions have been pivotal in advancing post-quantum cryptography?

MC: As we navigate the complex landscape of quantum-safe applications and networks, it's clear that our industry's response requires a collaborative approach. This is not a challenge that can be solved by one firm or organization alone. It requires specialized expertise, innovation, agility, and a strong focus on customer intimacy.

Collaboration is vital -- working together to achieve a common goal. Nokia and our collaborators are engaging and bringing together the best minds and expertise from across the quantum and security industry to drive innovation and progress. We are engaged in partnerships with QKD experts, and Public Key Infrastructure with Post-Quantum Cryptography (PKI-PQC) specialists and more.

Also: Why you should power off your phone once a week -.

Using a unified language and framework can help raise awareness about the threat of quantum attacks and the solution of quantum-safe networks. But it's not just about language -- it's about action. We need collaboration across various players, including application providers, technology vendors, system integrators, research institutions, connectivity providers, and quantum technology innovators.

By working together, we can drive progress, innovation, and adoption of quantum-safe networks. Ultimately, Nokia can ensure that our clients and industries are protected from the threats of the evolving quantum threat landscape.

ZDNET: What would you say to organizations that feel the quantum threat is too distant to warrant immediate action?

MC: While a CRQC may not exist yet, investment and technological evolution are continuing at an accelerating pace, with experts predicting that a CRQC will be available within the next 5 to 15 years. Transitioning systems takes time; therefore, it's crucial to act now to mitigate your future risks.

Furthermore, encrypted data can be harvested today and held to be decrypted later when CRQCs become accessible, a strategy known as "harvest now, decrypt later" (HNDL). By implementing quantum-safe measures now, consumers can protect their data's integrity, confidentiality, and authenticity today and for the quantum future.

Lastly, everyone should understand that the whole ICT sector is migrating to new quantum-safe cryptography. Thus, immediate action should take place for an organization to plan, define, and execute an ordered and resilient migration. Such an approach will minimize risk and costs.

ZDNET: Could you share your vision of what a fully quantum-safe critical infrastructure might look like in the next 10–20 years?

MC: In the next 10 to 20 years, we foresee a fully quantum-safe digital world, where advanced quantum-safe technologies will protect sensitive data at both the application and network layers. Post-Quantum Cryptography (PQC), Pre-Shared Key (PSK) cryptography, and Quantum Key Distribution (QKD) will ensure secure, confidential, and tamper-proof communications.

We believe this world will be built on a robust defense-in-depth framework, ensuring that the entire communication fabric is quantum-secure against both current quantum threats and future advancements in code-breaking.

This will be realized by complementing quantum-safe applications with network-level quantum-safe cryptography, embracing a crypto-resilient approach that utilizes both asymmetric and symmetric cryptography.

In this future world, organizations will employ AI-driven risk assessment tools to continuously monitor and mitigate potential quantum threats. This will ensure that security, privacy, and trust -- essential elements for our digital economies -- create a robust, crypto-resilient world capable of withstanding the challenges posed by quantum computing.

That presented, let's remember that this vision of a quantum-safe future begins now, today, safeguarding generations to come.

ZDNET: Lastly, how do you foresee quantum-safe encryption evolving as quantum computing technologies mature?

MC: Depending on the timeframe, as we advance with quantum communication, the pure act of connecting to one another will need to be quantum-safe. All communications will need to be quantum-safe.

As the world moves forward and technology evolves, the threats will similarly evolve. So, much like our world today, we will need to continue to stay on top of emerging threats. Unfortunately, no silver bullet will solve all of our cybersecurity challenges.

It's an arms race of sorts, but there are powerful tools that can be deployed in a proactive way to mitigate the risk to our economy and society.

Quantum computing is on the horizon, and its impact on cybersecurity, encryption, and digital infrastructure is becoming increasingly urgent. How concerned are you about the potential risks of quantum-based cyberattacks?

Have you or your organization started considering quantum-safe encryption solutions? Do you think governments and industries are moving quickly enough to address these challenges? What role do you think AI will play in either strengthening or weakening cybersecurity in a post-quantum world? Let us know in the comments below.

Can you jailbreak Anthropic's latest AI safety measure? Researchers want you to try -- and are offering up to $20,000 if you suc......

Metric collection is an essential part of every machine learning project, enabling us to track model performance and monitor training progress. Ideall......

When you think of headphones, AR/VR headsets, and smart jewelry companies OpenAI may not come to mind. However, a new tra......

OpenAI eyes the wearables business: Robots, headsets, watches and a whole lot more

When you think of headphones, AR/VR headsets, and smart jewelry companies OpenAI may not come to mind. However, a new trademark filing points to that might soon change.

OpenAI filed a trademark application on January 31 with the US Patent and Trademark Office (USPTO), spotted by TechCrunch. The application includes an extensive list of software offerings, including cloud computing, machine learning, and software-as-a-service (SaaS) -- all of which are not unexpected for a organization like OpenAI. However, sprinkled within the list are a significant number of hardware products.

Also: ChatGPT's Deep Research just identified 20 jobs it will replace. Is yours on the list?

The hardware products include earphones, headphones, sunglasses, smart watches, smart bands, smart jewelry, wearable computers, wearable cameras, digital media streaming devices, virtual and augmented reality headsets, goggles, glasses, controllers, and remotes.

The common thread among these products is that they are all wearables, which align perfectly with OpenAI's current offerings. One thing that large language models (LLMs) can do very well is synthesize robust amounts of data and draw insights. Then those insights can be shared in a conversational matter and formatted to be digestible.

For that reason, many AI wellness wearables on the market are already incorporating AI within their platforms. For example, the Whoop band has an OpenAI-powered Whoop Coach, a conversational chatbot you can use to learn more about your health data and life habits. Ultimately, LLMs help clients make sense of the expansive data their wearables collect.

Also: You can access ChatGPT Search without an account now - here's how.

Smart sunglasses and AR/VR headsets are another natural space for OpenAI to explore, as these products have a heavy AI presence. For example, the Meta Rayban AI and Meta Quest 3 headset incorporate Meta AI to assist you with everyday tasks, the way a more advanced AI assistant would.

Other new products, such as the Halliday and Even Realities sunglasses, incorporate AI to provide similar advanced assistance, such as transcribing and translating audio in real-time. Beyond voice assistants, machine learning techniques play a role in training and enhancing the performance of AR/VR headsets for computer vision and object recognition.

Another interesting item on the list was the reference to humanoid robots. The study specifically mentions, "User-programmable humanoid robots, not configured," and "Humanoid robots having communication and learning functions for assisting and entertaining people."

This mention of humanoid robots aligns with OpenAI's new job listings on its robotics team made earlier this month. The positions include EE Sensing Engineer, Robotics Mechanical Design Engineer, and TPM Manager.

Also: Wanted: Humans to build robots for OpenAI - and not everyone is thrilled.

At the time, Caitlin Kalinowski, who joined OpenAI in November to lead the robotics and consumer hardware team, noted it was yet to be determined whether the focus was to be humanoid robots. She also added, in reply to a user concerned about OpenAI's entry into hardware, "Imagine being able to send a robot to do a job that's really dangerous for a human."

Just yesterday, Figure AI, the robotics organization, showcased it was ending its partnership with OpenAI to focus on building its own in-house AI. This could now lead OpenAI to build its own robots in-house.

OpenAI has grown ChatGPT into many different offerings and products, including ChatGPT for Enterprise, Gov and Edu, ChatGPT Search, and more. Hardware is the logical next frontier for the AI research corporation. However, that doesn't mean an OpenAI hardware offering will be on the market any time soon.

It's common practice for companies to make these types of filings while exploring new ideas; often, such projects never see the light of day. Even if any of these projects do come to fruition, it could be a long roadmap before any are available to consumers.

What you can expect soon is for OpenAI to continue expanding its AI software offerings, particularly in ChatGPT. In the past week alome, OpenAI unveiled its o3-mini model and Deep Research -- each one advancing the degree of assistance individuals can get from the chatbot.

Metric collection is an essential part of every machine learning project, enabling us to track model performance and monitor training progress. Ideall......

Remember Nokia? Back before smartphones, many of us carried Nokia's nearly indestructible cell phones. They no longer make p......

In October 2023, former president Joe Biden signed an executive order that included several measures for regulating AI. On his ......

As AI Agents Evolve, So Does Their Definition

The AI industry is accelerating rapidly, and this is evident in the introduction and application of AI agents. A few years back, AI was just an LLM wrapper, but now, with options like Operator by OpenAI, it is much broader.

At MLDS 2025, India’s largest GenAI summit for developers, organised by AIM, Siddhant Goswami, co-founder of 100xEngineers, an AI lab specialising in Generative AI, explained how AI agents have evolved and changed over the past few years.

Goswami noted that over the past two years, ever since he started developing AI Agents, their definition has changed significantly each year and each quarter. He referenced “God In A Box,” which many had initially perceived as an AI agent, but it merely was an LLM wrapper. He emphasised that perspectives are shifting and that the entire industry is advancing rapidly.

Not just Goswami, but many Indian founders, in general, love AI agents.

The journey of AI agents can be divided into distinct phases, each introducing a new level of sophistication and functionality. Initially, everyone was clueless about why they existed, but they are now building meaningful solutions based on this knowledge.

Goswami outlines five phases of AI development. The first phase began with large language models (LLMs) like ChatGPT in 2022, which could process text inputs but lacked advanced reasoning. The second phase improved context windows, allowing people to input extensive texts, leading to the practical use of corporate databases.

The third phase introduced retrieval-augmented generation, enabling LLMs to access external knowledge. The fourth phase brought multimodal capabilities, allowing LLMs to process images, voice, and video. In the fifth phase, LLMs utilise memory layers, enhancing their ability to remember and adapt responses.

Building an AI Agent as an Artificial Being.

Summing up the phases of building an AI agent, Goswami told AIM, “We are kind of recreating a being, an artificial being. We first gave them hands and legs, then we taught them to learn the language, and then we gave them memories so they could remember things. I feel that’s the evolution of AGI itself.”.

With this type of advancement in AI agents, he believes we can expect to achieve AGI sooner.

Especially when Indian founders, in general, love AI agents.

At MLDS, Goswami mentioned that AI agents can hallucinate and go off the rails. So, a human-in-the-loop model is needed to monitor the operations.

“Even if you build the best technology right now, without having a human in the loop and giving feedback, I don’t think we would be able to achieve the level of intelligence we are expecting from this.”.

Despite the exciting progress, Goswami warned that not everything requires an AI agent. Many businesses rush into agent-based automation without understanding the trade-offs.

Most AI problems can be solved with a simple LLM and a RAG setup. Companies implementing AI solutions should prioritise simplicity. He suggested not to overcomplicate things and only introduce agents if necessary. If a task requires multiple steps, reasoning, and external tools, that is when we should consider using an AI agent.

It looks like AI agents will become increasingly autonomous, with longer memory retention and real-world action-taking abilities.

Research and Market’s study on ‘AI Agents Market Analysis projects the market for AI agents to grow from $[website] billion in 2024 to $[website] billion in 2030, with a CAGR of [website] during 2024-2030.

For now, with human observation added, the definition of AI agents should continue to evolve. It’s exciting to see what the future holds for them.

Security platform Astra Security raised $[website] million in a growth capital round led by Emergent Ventures. Neon Fund, more effective Capital, Blume Ventures, an...

Hallucinations continue to be one of the most critical challenges with AI. Although databases and a few other methods can mitigate this, they’re not t...

The last day of MLDS 2025, India’s largest developer conference, kicked off with Paras Chopra, founder of Turing’s Dream (now LossFunk), discussing ho...