Chatbots that never say die

Chatbots that never say die


Recently, when software engineer Blake Lemoine interviewed Google’s chatbot LaMDA, its responses were so real that it convinced many that it was sentient.

Artificial intelligence-based Large Language Models (LLMs) software can take in a few sentences and come up with convincing replies.

Last year, San Francisco Chronicle carried a story (recalled in the recent  MIT Technology Review) about a person who uploaded old texts and Facebook messages from his deceased fiancée and created a chatbot version of her. It reportedly gave him a lot of peace.

Artificial intelligence has made it possible to mimic voices, called ‘voice cloning’. In June, Amazon shared an audio of a little boy listening to a passage from  The Wizard of Oz, read by his recently deceased grandmother.

“Her voice was artificially re-created using a clip of her speaking that lasted for less than a minute,” says  MIT Technology Review.

From a reading dead grandmother to a talking dead grandmother is but a small leap. So, you can ‘talk to the dead’.

Chatbots aren’t sentient but they can pretend to be if fed with lots and lots of data. In today’s world that shouldn’t be a problem. Only, things could swing to the other extreme — two chatbots in an unending conversation with each other.





Source link

Drones, robots and other gizmos with green thumb

Drones, robots and other gizmos with green thumb


The technological leap in agriculture from chemical fertilisers and pesticides to smartphone-based remote operation of irrigation motors happened about a decade ago. Over the last few years, agricultural fields have been awash with more technology.

Today, tech-savvy farmers use drones to identify and root out infected plants before they can damage the entire crop. This solution, of course, is not affordable to all farmers.

But into the wide gap between smartphone-based motor switches and drones and robots, more technology is rushing in. For example, weather stations are being put up on agricultural fields.

Frugal engineering

A weather station at a farm near Erode, Tamil Nadu

In Perundurai, a remote part of Tamil Nadu, a small-sized company called Mobitech Wireless Solutions has been selling products like valves, and wired and wireless valve controllers and sensors. These enable farmers to operate their farm equipment from home, using smartphones. SIM cards inserted in motors or valves do the trick. Using the smartphone’s timer, the farmer can irrigate his fields for fixed periods. Yet another layer of automation is the use of ‘internet of things’ or IoT — the sensors tell the controller when to switch on the irrigation motors and how much water to use. Company officials tell  Quantum that business has been growing 40 per cent annually in the last few years, thanks partly to the pandemic and the resultant labour shortage. These IoT-connected sensors typically save 35 per cent in water consumption, the company says.

And now, Mobitech Wireless has launched a weather station — its priciest product at ₹25,000 each. How does the weather station aid the farmer? While other products give the farmer control over irrigation, the knowledge of wind speeds and outside temperature can help optimise water use, explains P Dhanasekaran, General Manager, Mobitech Wireless. For instance, the sensors can gauge soil moisture levels but the weather station could give the controller a word of caution: don’t let out water now, because the wind speeds are high.

An added advantage for farmers, Dhanasekaran said, is that the data proves handy when lodging an insurance claim following crop damage due to bad weather. This saves time and effort for the farmer, who would otherwise be forced to get the data from the district weather station.

Further automation

Beyond sensor-IoT-based agriculture, you have ‘deep learning’ (a subset of machine learning) and fuzzy logic lending a hand in crop management. In a recent scientific paper published in  Smart Agriculture Technology journal, researchers Anis Ahmed, Dharmendra Saraswat and Aly El Gamal of Purdue University note that “the use of deep learning techniques for plant disease diagnosis provides multiple advantages, including separating disease symptoms, identifying multiple diseases and estimating disease severity”. They point out that deep learning techniques have been used on images acquired from multispectral and hyperspectral sensors to identify the disease-affected regions in a field; thermal sensors and fluorescent sensors have also been used.

Plant disease identification is necessary for precision agriculture and plant phenotyping. Both these areas are data-, information-, and technology-intensive. But researchers today have access to data, mainly from different collection platforms, handheld sensors and drones. Deep learning has become a preferred approach for disease identification due to increased computational power, storage capacities, and availability of large datasets, the authors say.

Another paper, authored by Kalavathi Devi Thangavel, et al, of Kongu Engineering College, Perundurai, Erode, focuses on the specific design of intelligent microclimate controllers for greenhouses.

Prof Thangavel told  Quantum that their work was based on computer simulations of data obtained from a local greenhouse. The simulation took temperature, humidity and soil moisture as inputs and regulated the operations of the heater, pump and humidifier. This, of course, remains to be tested in a real greenhouse, but it is believed that high crop yields are possible.

All these indicate that the world is just beginning to scratch the surface of emerging technologies for agriculture.





Source link

How to jettison fossil fuel

How to jettison fossil fuel


Flying is an activity that emits greenhouse gases, and the main culprit is aviation turbine fuel — a kerosene. Therefore, a global search is on to replace kerosene, at least partly, with a more environment-friendly fuel. The search has yielded a class of plant-based fuels that have now come to be known as ‘drop-in sustainable aviation fuels’ or SAF, where ‘drop-in’ refers to functional similarity to fossil fuels. As Aaditya Khanal and Mohammad Shahriar of University of Texas, Tyler, observe, the carbon dioxide production during SAF combustion in aircraft engines “is roughly equivalent to the carbon dioxide absorbed by the plants to produce the biomass”. So, we have a solution at hand, right? Wrong.

The problem

The problem is to do with economics. SAF has been known about for years. The first commercial flight with 50 per cent SAF blend was KLM’s Boeing 737-400 between Amsterdam and Paris, carrying 171 passengers, on June 29, 2011. Since then, according to Khanal and Shahriar, over 2,500 commercial passenger flights of 22 different airlines have used 50 per cent SAF derived from jatropha, cooking oil, camelina and sugar cane. Yet, the consumption of SAF is less than a per cent of total jet fuel.

Average global SAF production from 2013 to 2015 was 0.29 million litres per year, which rose to 6.45 million litres per year from 2016 to 2018. Additionally, annual global SAF production was projected to reach 8 billion litres by 2032. Sounds like much, but this is really like offering a banana to a hungry elephant.

The first-generation SAF is useless because it is made from edible oils needed for food. The second-generation SAF — produced from jatropha, castor, pongamia pinnata and so on — is the one under consideration, because the third-generation SAF, produced from photosynthetic algae, emits more greenhouse gases than it saves.

The conventional process for producing second-generation SAF is known as HEFA ( hydroprocessed esters and fatty acids), which calls for the removal of oxygen present in fatty acids in plant oils, by adding hydrogen. Now you know the problem — hydrogen. The process needs energy and one must also consider the land-use change needed to grow crops for SAF, and the water and fertiliser consumption. Overall, SAF is two to five times costlier than conventional jet fuel.

The solution

In a research paper published in the preprint server bioRxiv.org, Timothy Sheppard, et al, suggest a new technology called ‘electromicrobial production’ (EMP) of SAF, which, as the name suggests, uses microbes. “Production of hydrocarbons using electrically powered microbes employing fatty acid synthesis-based production of alkanes could be an efficient means to produce drop-in replacement jet fuels using renewable energy,” the authors say. These microbes “have an extraordinary ability to manufacture organic compounds using electricity as the primary source of metabolic energy”, they say. This process uses light, atmospheric carbon dioxide and electricity.

Traditionally, engineered cyanobacteria are used for microbial production, but they are difficult to engineer. Sheppard points to a better microbe,  Vibrio natriegens, capable of ‘extracellular electron uptake’ (EEU).

There are two ways of getting microbes to produce biomass. One is hydrogen oxidation, where the microbes consume hydrogen to produce biomass. The second — EEU — involves delivery of electrons into cells, either through a diffusible intermediary such as water-soluble quinones, or through direct electrical contact with an anode.

“We believe the time is right to start scaling up production of jet fuels with EMP,” say the authors of the paper. They believe that “hydrogen-mediated EMP” is a slightly more efficient method, but are also working on EEU as a viable alternative.

Peer comment

Asked for a response to the paper, Dr Anjan Ray, Director, Indian Institute of Petroleum, Dehradun (which is also engaged in the development of SAF) told  Quantum: “Conceptually, it is rather exciting to imagine an electromicrobial system, as described by the authors, with the theoretical conversion efficiencies indicated.”

However, Dr Ray cautioned against undue optimism. The results in the paper indicate only the theoretical possibilities, not the practical limitations, of the proposed process of carbon fixation, wherein carbon dioxide provides the carbon source and electrical energy provides the metabolic energy for the conversion of carbon dioxide, by the microbe, into alkanes and terpenoids in the desired jet fuel range, he said.

Observing that Sheppard’s paper describes the best-case scenarios, which are impressive, Dr Ray said that the probability of practically achieving such efficiencies is not evident at this time. “I expect this to be a long haul of several years, if not decades,” he said.

He further observed that the “efficiencies only indicate the extent of energy conversion, not the kinetics (the rate at which such conversion happens over time)”. Electromicrobiological kinetics can vary widely, so a lot more research would be needed before one can be sure of producing an adequate volume of fuel.

“In essence, it is too early to comment on the chances of commercial success for this route or a timeline for such success — but the proposed pathway is potentially exciting and disruptive,” Dr Ray said.





Source link

Methane by industry, for industry

Methane by industry, for industry


With the next global climate conference (COP-27) barely 45 days away, there is heightened focus on how to avoid greenhouse gas emissions. Methane is a deadly greenhouse gas (though shorter-lived than carbon dioxide) and it is impossible to avoid production of methane altogether. For example, methane is produced in the oil industry when gas is flared.

One good way to neutralise the methane menace is to use the gas in applications where it is broken into its constituent elements. For this, we turn to an old industrial process — diamond coating.

One of the activities of the Materials Research Lab at IIT-Madras is to develop different technologies for coating surfaces with diamond to impart surface hardness, wear-resistance and lubrication properties for various applications. The developed technologies are commercialised by Kapindra Precision Engineering, a start-up incubated at IIT-Madras Research Park.

Prof MS Ramachandra Rao of the Department of Physics, IIT-M, told  Quantum that the process of diamond coating surfaces is a good way of using methane harmlessly.

But first, a cautionary note — there is no diamond here. Nor is there a sparkle, for it is all black carbon. Diamond is just one of the avatars of carbon, like coal and graphite. They all differ in the manner in which carbon atoms are structured, which in turn comes from how the electrons are arranged (in sub-orbitals).

Therefore, all you need to get a diamond coat is some carbon. A good provider of carbon is methane, which is a molecule of one carbon atom attached to four hydrogen atoms. To separate carbon from this molecule you need extremely high energy. The machines used for coating materials (such as tools or auto components) feature several extremely thin tungsten filaments. When electricity is passed through the filaments they become very, very hot — 2,200 degrees C. If you push methane through these hot filaments, the gas breaks, step-by-step, into hydrogen and carbon. The carbon deposits on any substrate kept below the filaments, forming a diamond-hard coat.

Carbon exists with three types of electron bonds — sp1, sp2 and sp3. These refer to the regions where the electrons are most likely found in each orbit around the nucleus. Basically, sp2 is graphitic carbon, while sp3 is diamond.

When methane gas is passed through the ultra-hot filaments and the carbon and hydrogen split, the carbon comes in both sp2 and sp3 forms. However, some of the hydrogen (just divorced from carbon) still has an affinity for sp2 carbon — it reunites with sp2 carbon, leaving behind sp3 carbon, or diamond, which is coated on the surface. Interestingly, the process also yields some pure hydrogen.

Diamond coating, thus, presents an opportunity to use methane and produce hydrogen, too.

The machines used for diamond coating are imported, but IIT-M has got one indigenously produced, at a fraction of the import cost.





Source link

Cactus packaging for food

Cactus packaging for food


Scientists at IIT-Roorkee have developed a biodegradable, antimicrobial packaging material by combining gelatin with two cacti varieties,  Cylindropuntia fulgida (CF) and  Euphorbia caducifolia extract (ECE), though the latter was used less. Gelatin is a good antimicrobial packaging material but has poor mechanical properties (like strength) and absorbs water. Adding the cacti countered these two issues.

In a paper on the material, published in the  International Journal of Biological Macromolecules, the authors — Lokesh Kumar, Ram Kumar Deshmukh and Kirtiraj Gaikwad — say they chose  Cylindropuntia fulgida as a polymer because of “ its low cost and availability”.

Prof Gaikwad told  Quantum that the gelatin-CF-ECE composite film is suitable for food packaging, given that all the ingredients are natural.

“The packaging film also exhibits antimicrobial properties, which make it suitable for the preservation of perishable fruits and vegetables. The composite film is flexible with excellent heat-sealing properties, can be converted into shelf-standing pouch for the packaging of low-moisture food products,” Gaikwad said.

Cerium oxide for bone health

Surgical cotton microfibres loaded with nanoceria (cerium oxide) could be a new platform for bone tissue engineering, according to a joint study by IIT-Roorkee and Indian Institute of Engineering Science and Technology (IIEST), Shibpur.

Bone regeneration is hampered by ‘oxidative stress’ — a situation in which cell-damaging ‘reactive oxygen species’ such as peroxides and superoxides exceed the anti-oxidants that can neutralise them. Polymer scaffolds loaded with material that scavenge these free radicals have always been used. Now, the IIT-Roorkee and IIEST study has discovered that integrating nanoceria into cellulose-gelatin and freeze-drying the mix to produce ‘CG-NC scaffolds’ helps a lot in bone regeneration.

“Adding nanoceria to the scaffolds improved mechanical, bio-mineralisation properties, and decreased swelling and in-vitro weight loss. In-vitro studies confirmed that CG-NC scaffolds supported cell proliferation and differentiation better than bare (CG) scaffold,” says a paper produced by the scientists and published in the  Ceramics International journal.

Fresh juice, forward osmosis

Forward osmosis is a good method for concentrating pomegranate juice with minimum effect on quality and extended shelf-life, a study by the Central Food Technological Research Institute, Mysore, has revealed.

Researchers Das Trishitman, Pradeep Negi and Navin Rastogi concentrated pomegranate juice using forward osmosis and thermal evaporation methods.

Based on the hydroxymethyl furfural content (less than 25 mg per kg), it was concluded that juices concentrated through forward osmosis could be stored at ambient and accelerated conditions for 101 and 66 days, respectively.

Comparatively, thermally concentrated juice could be stored only for 31 and three days, respectively. Further, forward osmosis also resulted in a four-fold increase in brix as well as anthocyanin content, says a paper produced by the scientists and published in the Food Chemistry journal.





Source link

Honey, who shrunk our brains?

Honey, who shrunk our brains?


Joe Bauers was put to sleep. When he woke up 500 years later, in a ‘human hibernation’ experiment that went bad, he found himself to be the smartest person in the whole world.

The message embedded in this storyline of the 2006 movie  Idiocracy is that humans are progressively getting dumber. The idea presumably arose from some research findings that the size of the human brain has been shrinking over millions of years, though there is no evidence that this has diminished man’s cognitive ability — which has, in fact, only increased.

The ‘shrinking brain’ theory was born out of studies of the skull size of Homininis — our ancestors. It has many takers. John Hawks of the University of Wisconsin long held this view, as did Christopher Stringer, a paleoanthropologist at the Natural History Museum, London. And, more recently, Jeremy DeSilva, paleoanthropologist at Dartmouth College, produced a paper, after studying 987 skulls, that dramatically announced that the brain size of human ancestors increased 2.1-1.5 million years ago, but started to sharply decrease 3,000 years ago, and is now a lemon smaller. This claim has been refuted by other scientists, notably Brian Villmoare, anthropologist at the University of Nevada, Las Vegas.

The general explanation is that, after humans became domesticated, turned agriculturists and learned to read and write, a big brain was not necessary. With gadgets assisting us, our brains could become even smaller.

Moral of story: if you want to be the smartest person on the planet, just sleep off for 500 years.





Source link

YouTube
Instagram
WhatsApp