S.D.I. English Edition newsletter: Trick or AI Treat … ?

The hot topic of the moment, AI; artificial intelligence today seems to be able to do almost anything, even coffee; is this a momentary fad or are we facing a step change in the everyday technological world?

Generative AI has entered everyday life, ChatGPT being the best known example, which has shown its capabilities, and its limitations; but it has also revealed the way to new and possible futuristic scenarios, but what can we expect in the coming times?

Scenarios and potential

Certainly in the panorama of artificial intelligence a further step has been taken, I am talking about image generation; what is looming as a rapidly evolving sector, and OpenAI, with its latest model DALL-E 3, seems to have reached a new technological apex, shifting the boundaries between textual description and visual representation. The ability to transform verbal descriptions into coherent and detailed visual images represents a quantum leap in the field of AI image synthesis.

The DALL-E 3 model, an evolution of its predecessors, is distinguished by its ability to create images by precisely following complex descriptions and managing the generation of text within the images, a capability that was difficult (not to say impossible) to achieve in previous models.

Microsoft has introduced DALL-E 3 into its Bing Chat and Bing Image Creator services, making the model freely available to a wider audience, hence the availability of DALL-E 3 to the general public and its integration into accessible platforms such as Bing Chat open up interesting but also complex scenarios in terms of demand management, security and ethical issues.

Other outlets for AI? In the cyber sec: monitoring the anomalous

Artificial intelligence (AI) is revolutionising many sectors, and cybersecurity is no exception. Cyber attacks are becoming increasingly sophisticated and frequent, and organisations are constantly looking for advanced methods to protect their digital assets. In the security field, the outlets are promising, both for the good guys and the bad guys.

For the good guys to protect themselves from cyber criminals, artificial intelligence comes to the rescue in several ways. Having a tool that can understand what is normal, and warn us when something out of the ordinary happens, is a great help in identifying attack attempts. Without AI, it is the human being who has to decide in advance which events should raise an alarm.

Deviation from the standard can be identified autonomously by using AI to detect suspicious behaviour in a file. By detecting threats, AI systems can analyse large volumes of data in real time to identify unusual patterns or suspicious activities. This makes the detection of new types of malware or intrusion attempts much faster and more efficient.

Machine learning offers AI models that can be trained to identify new types of attacks by analysing past trends. For instance, they can be trained to recognise phishing attempts based on previous emails.

Without going into specifics, we can say that there are machine learning algorithms designed for specific tasks, such as identifying an anomaly with respect to a baseline.

This is still something that humans do better, but not in real time and not on a large scale. Finally, extracting information from large data sets is something that a machine learning algorithm can do better than a human being.

Out of the standard boxes ...

AI can analyse user behaviour within a network and identify abnormal behaviour, such as attempts to access unauthorised resources, the system analyses a user's habitual behaviour to determine if access is legitimate. Once a threat is identified, AI systems can take automatic measures to isolate or neutralise the threat, reducing the time between detection and response.

But in addition to managing, we must prevent threats by using AI to predict where and when future attacks might occur, based on trends and historical data. Or by analysing possible vulnerabilities, scanning and analysing systems and software for potential vulnerabilities, helping organisations fix problems before they become an entry point for attackers.

Reducing false positives: One of the challenges in security is distinguishing between legitimate and potentially malicious activities. AI can help reduce the number of false alarms, making the real ones more obvious and less costly to manage.

From the perspective of the malicious ...

It is important to note that while AI offers many benefits in terms of cybersecurity, it also presents challenges.

Villains will have the opportunity to increasingly improve attack strategies, translations of texts used in phishing attacks, harnessing the power of AI engines in organised groups to find 'holes' in systems identified as potential victims.

Attackers can use AI to develop more sophisticated malware or to automate attacks. Moreover, as with any technology, the effectiveness of AI depends on the quality of its implementation and the state of the art of research.

In summary, AI has great potential in cybersecurity, but it is essential that organisations implement it in a prudent and informed manner.

Microsoft's next scenario

In the next few days we are expecting a big package to be released by Microsoft, Copilot promises spice and ubiquity within the Office suite but not only that, we have seen that Edge has already taken advantage of it for Dall-E-3 but recently it also has a prompt like interface with CoPilot and that's not all.

There will be 2 classes, standard free for all users and an enterprise class control logic for corporate users.

To sum up: Microsoft Copilot and Bing Chat remain free, but to have control of the data and sources we will have to add the Bing Chat Enterprise licence to our corporate tenant (licence holder).

What to do with Copilot

Copilot will be integrated and work alongside the classic Office packages; Word, Excel, Teams, Outlook, e.g. to summarise key points and action items in Microsoft Teams, draft new documents in Word, respond automatically in Outlook, create presentations in PowerPoint and more.

Not least, the D365 Suite will integrate new functions in all areas of Sales, Marketing, Customer Service; aimed at helping in everyday life and speeding up those tedious and repetitive tasks that we so detest.

Schedule meetings by automatically checking the availability of guests, generate a resume of the meeting at the end by adding data discussed and reports seen in the meeting

Integrated Customer Services for a better customer experience and enable agents to provide enhanced customer support,

For marketing new tools to classify customers, products, markets, brand penetration segments in sales scenarios.

Integrations with the retail world to speed up Time To Market, synthesising products and proposing accessories based on customer preferences. And for Developers?

The Power APP suite will be integrated with AI to allow both developers and non-technical end-users to create business apps by simply describing them.

Think PW App by leaving the tedium of writing to the generative CoPilot interface.

Beyond Microsoft ...

Obviously Microsoft's will not be the only ERP to benefit from AI innovations, for example SAP is integrating Joule Copilot into its ecosystem by integrating it into Sales, Marketing, Recruiting, Supply Chain processes and adapting it to the B2B and B2C world, procurement, etc.

For its part, Infor, increasingly better positioned in the Gartner quadrants; is preparing its own AI engine called Coleman.

Each player presents its own solution, the common denominator being to collaborate with users to relieve them of work repetitiveness or the complexity of statistical calculations impossible without the use of dedicated GPUs.

The ability to make decisions? Let's leave it for now to humans, with their ability to create, improvise and even make mistakes.

Conclusions

Are we facing a new industrial revolution or just a trick for Halloween?

I lean towards the former, although it would be plausible to have an EU investment policy to help companies that invest in this direction, the returns in terms of jobs and strategic positioning would certainly be significant.

This, of course, requires heavy investment, which is currently low in Europe and conspicuous in China and the US. Co-pilots will increasingly enter the operational life of our days, helping and supporting us in our work. We will get our first taste on 1/11/23 with the official release of the Microsoft suite, but more news are on the way; waiting for Gemini from Google ...

Let's expect incremental developments of solutions, the technological aggregation of solutions could give a breakthrough and a spin to solutions, see Infor's proposal, Robotic Process Automation integrating Rpa capabilities, Machine learning with AI.

But what about privacy regulations?

Much smoke and little fire, at the moment there is no precise and clear governance for protection, and legal safeguards are needed on this too; Microsoft Chat will try to do this at least by protecting corporate data, think of the dissemination of patents or impactful projects spread without consent.

On this, too, AI promises battle, should we protect our data? Yes, no, maybe

Thanks to Paolo Lanzani to share Infor platform articles.